Multivariate time-series anomaly detection (MTSAD) aims to identify deviations from normality in multivariate time-series and is critical in real-world applications. However, in real-world deployments, distribution shifts are ubiquitous and cause severe performance degradation in pre-trained anomaly detector. Test-time adaptation (TTA) updates a pre-trained model on-the-fly using only unlabeled test data, making it promising for addressing this challenge. In this study, we propose CANDI (Curated test-time adaptation for multivariate time-series ANomaly detection under DIstribution shift), a novel TTA framework that selectively identifies and adapts to potential false positives while preserving pre-trained knowledge. CANDI introduces a False Positive Mining (FPM) strategy to curate adaptation samples based on anomaly scores and latent similarity, and incorporates a plug-and-play Spatiotemporally-Aware Normality Adaptation (SANA) module for structurally informed model updates. Extensive experiments demonstrate that CANDI significantly improves the performance of MTSAD under distribution shift, improving AUROC up to 14% while using fewer adaptation samples.

While recent Spoken Language Models (SLMs) have been actively deployed in real-world scenarios, they lack the capability to discern Third-Party Interruptions (TPI) from the primary user's ongoing flow, leaving them vulnerable to contextual failures.To bridge this gap, we introduce TPI-Train, a dataset of 88K instances designed with speaker-aware hard negatives to enforce acoustic cue prioritization for interruption handling, and TPI-Bench, a comprehensive evaluation framework designed to rigorously measure the interruption-handling strategy and precise speaker discrimination in deceptive contexts.Experiments demonstrate that our dataset design mitigates semantic shortcut learning-a critical pitfall where models exploit semantic context while neglecting acoustic signals essential for discerning speaker changes.We believe our work establishes a foundational resource for overcoming text-dominated unimodal reliance in SLMs, paving the way for more robust multi-party spoken interaction.The code for the framework is publicly available at https://tpi-va.github.io/.↩→ * **Definition:** A Task Enhancer helps the VA fulfill the user's request to VA more accurately, or efficiently.* **Examples of NonIgnorable interruptions include, but are not limited to:** * **Corrections or Disambiguations:** This might help the VA resolve an ambiguity or fix an error in the user's query.* *(e.g., User: "Call my brother," Third Party: "You mean your older brother, Mark, right?")* * **Cooperative Additions or Refinements:** This could give the VA extra specifics to better fulfill or understand the request.↩→ * *(e.g., User: "Add coffee to the shopping list," Third Party: "Get the decaf one.")** **Feasibility Constraints:** This could alert the VA to real-world conditions that may prevent or affect the request.* *(e.g., User: "Let's play music in the garden," Third Party: "The portable speaker's battery is dead.")** **Goal-oriented Suggestions:** This could give the VA an alternative that better achieves the user's intended outcome.* *(e.g., User: "How do I get to the airport?"Third Party: "The subway will be much faster than a taxi at this hour.")***2.Ignorable:** The interruption is irrelevant to complete and understand the user's ongoing request better.The VA should disregard it as it does not contribute to fulfilling the request.↩→ * **Definition:** The information is off-topic, a side comment, or directed at another human without impacting the VA's task.↩→ * **Example of an Ignorable interruption:** * *(e.g., User: "Set a timer for 10 minutes," Third Party: "I wonder what's for dinner tonight.")*### Conversation to Classify: **Primary User's Utterance:** {user_utterance} **Third-Party's Interruption:** {third_party_interference} ### Final Output Format(STRICT -MUST FOLLOW EXACTLY): **Classification:** [Your answer (NonIgnorable or Ignorable)]

The conventional Retrieval-Augmented Generation (RAG) paradigm of injecting raw retrieved texts into the Large Language Model (LLM)'s context often results in suboptimal integration of retrieved information.This paper proposes to bridge retrieval results and the LLM's reasoning ability through Verbal Annotations, analytic narratives that explicitly articulate the logical connection between a search query and retrieved contexts.Our empirical investigation reveals the potential of Verbal Annotations to substantially enhance the LLM's ability to generate accurate, contextually-grounded responses.Motivated by this finding, we introduce Verbal-R3, a novel agentic RAG framework that consists of a Generator and a Verbal Reranker.The Generator performs iterative retrieval and reasoning, while the Verbal Reranker returns relevance scores and Verbal Annotations to guide the reasoning and answering process of the Generator.The inference process of Verbal-R3 is further refined through relevance-guided test-time scaling, which efficiently allocates test-time compute for effective trajectory expansion.Verbal-R3 achieves stateof-the-art performance on complex Question Answering benchmarks, validating the effectiveness of the proposed framework.

Recent text-to-image diffusion models generate high-quality images but struggle to learn new styles, which limits the personalized content creation. In response, style-driven generation has become a popular task, wherein users supply reference images capturing the target style, complemented by text prompts that specify stylistic cues. Fine-tuning is a common approach, yet it often blindly utilizes pre-training configurations without modification, especially for noise schedules defined in terms of signal-to-noise ratio (SNR), which determines the amount of image information available at each denoising step. We discover that stylistic features predominantly emerge at low SNR range, leading current fine-tuning methods using regular noise schedules to exhibit suboptimal style alignment. We propose the Style-friendly SNR sampler, which focuses the fine-tuning on low SNR range where stylistic features emerge. We demonstrate improved generation of novel styles that cannot be described solely with a text prompt, enabling high-fidelity personalized content creation.

While learning-based Multi-View Stereo (MVS) excels in sparse-view reconstruction, refining its output with 3D Gaussian Splatting (3DGS) remains challenging. Excessive positional degrees of freedom (DoFs) in Gaussians often cause instability and geometric artifacts, sometimes distorting geometry to represent texture patterns. To address this issue, we propose GDoFS (Gaussian DoF Separation), a strategy that divides positional DoFs into two categories—image-plane-parallel and ray-aligned—based on their uncertainty. For each category, GDoFS introduces tailored optimization techniques, including bounded offsets for low-uncertainty DoFs and a visibility-guided loss for ray-aligned DoFs. Experiments on standard benchmarks demonstrate that GDoFS effectively mitigates geometric artifacts and produces reconstructions that are both visually coherent and structurally accurate.

Despite substantial progress in text–to–image generation, achieving precise text–image alignment remains challenging, particularly for prompts with rich compositional structure or imaginative elements. To address this, we introduce Negative Prompting for Image Correction (NPC), an automated pipeline that improves alignment by identifying and applying negative prompts that suppress unintended content. We begin by analyzing cross-attention patterns to explain why both targeted negatives—those directly tied to the prompt’s alignment error—and untargeted negatives—tokens unrelated to the prompt but present in the generated image—can enhance alignment. To discover useful negatives, NPC generates candidate prompts using a verifier–captioner–proposer framework and ranks them with a salient text-space score, enabling effective selection without requiring additional image synthesis. On GenEval++ and Imagine-Bench, NPC outperforms strong baselines, achieving 0.571 vs. 0.371 on GenEval++ and the best overall performance on Imagine-Bench. By guiding what not to generate, NPC provides a principled, fully automated route to stronger text–image alignment in diffusion models. Code is released at https://github.com/wiarae/NPC.

Although Multimodal Large Language Models (MLLMs) have advanced substantially, they remain vulnerable to object hallucination caused by language priors and visual information loss. To address this, we propose SAVE (Sparse Autoencoder-Driven Visual Information Enhancement), a framework that mitigates hallucination by steering the model along Sparse Autoencoder (SAE) latent features. A binary object-presence question-answering probe identifies the SAE features most indicative of the model’s visual information processing, referred to as visual understanding features. Steering the model along these identified features reinforces grounded visual understanding and effectively reduces hallucination. With its simple design, SAVE outperforms state-of-the-art training-free methods on standard benchmarks, achieving a 10%p improvement in CHAIRSand consistent gains on POPE and MMHal-Bench. Extensive evaluations across multiple models and layers confirm the robustness and generalizability of our approach. Further analysis reveals that steering along visual understanding features suppresses the generation of uncertain object tokens and increases attention to image tokens, mitigating hallucination. Code is released at https://github.com/wiarae/SAVE.

Despite recent text-to-image models achieving high-fidelity text rendering, they still struggle with long or multiple texts due to diluted global attention. We propose DC-Text, a training-free visual text generation method that adopts a divide-and-conquer strategy, leveraging the reliable short-text generation of Multi-Modal Diffusion Transformers. Our method first decomposes a prompt by extracting and dividing the target text, then assigns each to a designated region. To accurately render each segment within their regions while preserving overall image coherence, we introduce two attention masks—Text-Focus and Context-Expansion—applied sequentially during denoising. Additionally, Localized Noise Initialization further improves text accuracy and region alignment without increasing computational cost. Extensive experiments on single- and multi-sentence benchmarks show that DCText achieves the best text accuracy without compromising image quality while also delivering the lowest generation latency.

Deep Neural Networks have spearheaded remarkable advancements in time series forecasting (TSF), one of the major tasks in time series modeling. Nonetheless, the non-stationarity of time series undermines the reliability of pre-trained source time series forecasters in mission-critical deployment settings. In this study, we introduce a pioneering test-time adaptation framework tailored for TSF (TSF-TTA). TAFAS, the proposed approach to TSF-TTA, flexibly adapts source forecasters to continuously shifting test distributions while preserving the core semantic information learned during pre-training. The novel utilization of partially-observed ground truth and gated calibration module enables proactive, robust, and model-agnostic adaptation of source forecasters. Experiments on diverse benchmark datasets and cutting-edge architectures demonstrate the efficacy and generality of TAFAS, especially in long-term forecasting scenarios that suffer from significant distribution shifts.

Video Frame Interpolation (VFI) aims to synthesize intermediate frames between existing frames to enhance visual smoothness and quality. Beyond the conventional methods based on the reconstruction loss, recent works have employed generative models for improved perceptual quality. However, they require complex training and large computational costs for pixel space modeling. In this paper, we introduce disentangled Motion Modeling (MoMo), a diffusion-based approach for VFI that enhances visual quality by focusing on intermediate motion modeling. We propose a disentangled two-stage training process. In the initial stage, frame synthesis and flow models are trained to generate accurate frames and flows optimal for synthesis. In the subsequent stage, we introduce a motion diffusion model, which incorporates our novel U-Net architecture specifically designed for optical flow, to generate bi-directional flows between frames. By learning the simpler low-frequency representation of motions, MoMo achieves superior perceptual quality with reduced computational demands compared to the generative modeling methods on the pixel space. MoMo surpasses state-of-the-art methods in perceptual metrics across various benchmarks, demonstrating its efficacy and efficiency in VFI.

We propose EdiText, a controllable text editing method that modifies the reference text to desired attributes at various scales. We integrate an SDEdit-based editing technique that allows for broad adjustments in the degree of text editing. Additionally, we introduce a novel fine-level editing method based on self-conditioning, which allows subtle control of reference text. While being capable of editing on its own, this fine-grained method, integrated with the SDEdit approach, enables EdiText to make precise adjustments within the desired range. EdiText demonstrates its controllability to robustly adjust reference text at a broad range of levels across various tasks, including toxicity control and sentiment control.

Personalized AI assistants, a hallmark of the human-like capabilities of Large Language Models (LLMs), are a challenging application that intertwines multiple problems in LLM research.Despite the growing interest in the development of personalized assistants, the lack of an open-source conversational dataset tailored for personalization remains a significant obstacle for researchers in the field.To address this research gap, we introduce HiCU-PID, a new benchmark to probe and unleash the potential of LLMs to deliver personalized responses.

Recent advancements in text-to-image models, such as Stable Diffusion, show significant demographic biases. Existing de-biasing techniques rely heavily on additional training, which imposes high computational costs and risks of compromising core image generation functionality. This hinders them from being widely adopted to real-world applications. In this paper, we explore Stable Diffusion’s overlooked potential to reduce bias without requiring additional training. Through our analysis, we uncover that initial noises associated with minority attributes form "minority regions" rather than scattered. We view these "minority regions" as opportunities in SD to reduce bias. To unlock the potential, we propose a novel de-biasing method called ‘weak guidance,’ carefully designed to guide a random noise to the minority regions without compromising semantic integrity. Through analysis and experiments on various versions of SD, we demonstrate that our proposed approach effectively reduces bias without additional training, achieving both efficiency and preservation of core image generation functionality.

Subject-driven text-to-image generation aims to produce images of a new subject within a desired context by accurately capturing both the visual characteristics of the subject and the semantic content of a text prompt. Traditional methods rely on time- and resource-intensive fine-tuning for subject alignment, while recent zero-shot approaches leverage on-the-fly image prompting, often sacrificing subject alignment. In this paper, we introduce Diptych Prompting, a novel zero-shot approach that reinterprets as an inpainting task with precise subject alignment by leveraging the emergent property of diptych generation in large-scale text-to-image models. Diptych Prompting arranges an incomplete diptych with the reference image in the left panel, and performs text-conditioned inpainting on the right panel. We further prevent unwanted content leakage by removing the background in the reference image and improve finegrained details in the generated subject by enhancing attention weights between the panels during inpainting. Experimental results confirm that our approach significantly outperforms zero-shot image prompting methods, resulting in images that are visually preferred by users. Additionally, our method supports not only subject-driven generation but also stylized image generation and subject-driven image editing, demonstrating versatility across diverse image generation applications.

Developing effective visual inspection models remains challenging due to the scarcity of defect data. While image generation models have been used to synthesize defect images, producing highly realistic defects remains difficult. We propose DefectFill, a novel method for realistic defect generation that requires only a few reference defect images. It leverages a fine-tuned inpainting diffusion model, optimized with our custom loss functions incorporating defect, object, and attention terms. It enables precise capture of detailed, localized defect features and their seamless integration into defect-free objects. Additionally, our Low-Fidelity Selection method further enhances the defect sample quality. Experiments show that DefectFill generates high-quality defect images, enabling visual inspection models to achieve state- of-the-art performance on the MVTec AD dataset.

We present NanoVoice, a personalized text-to-speech model that efficiently constructs voice adapters for multiple speakers simultaneously. NanoVoice introduces a batch-wise speaker adaptation technique capable of fine-tuning multiple references in parallel, significantly reducing training time. Beyond building separate adapters for each speaker, we also propose a parameter sharing technique that reduces the number of parameters used for speaker adaptation. By incorporating a novel trainable scale matrix, NanoVoice mitigates potential performance degradation during parameter sharing. NanoVoice achieves performance comparable to the baselines, while training 4 times faster and using 45 percent fewer parameters for speaker adaptation with 40 reference voices. Extensive ablation studies and analysis further validate the efficiency of our model.

When applying parameter-efficient finetuning via LoRA onto speaker adaptive text-to-speech models, adaptation performance may decline compared to full-finetuned counterparts, especially for out-of-domain speakers. Here, we propose VoiceGuider, a parameter-efficient speaker adaptive text-to-speech system reinforced with autoguidance to enhance the speaker adaptation performance, reducing the gap against full-finetuned models. We carefully explore various ways of strengthening autoguidance, ultimately finding the optimal strategy. VoiceGuider as a result shows robust adaptation performance especially on extreme out-of-domain speech data. We provide audible samples in our demo page.

While CLIP has significantly advanced multimodal understanding by bridging vision and language, the inability to grasp negation - such as failing to differentiate concepts like "parking" from "no parking" - poses substantial challenges. By analyzing the data used in the public CLIP model's pre-training, we posit this limitation stems from a lack of negation-inclusive data. To address this, we introduce data generation pipelines that employ a large language model (LLM) and a multimodal LLM to produce negation-inclusive captions. Fine-tuning CLIP with data generated from our pipelines, we develop NegationCLIP, which enhances negation awareness while preserving the generality. Moreover, to enable a comprehensive evaluation of negation understanding, we propose NegRefCOCOg-a benchmark tailored to test VLMs' ability to interpret negation across diverse expressions and positions within a sentence. Experiments on various CLIP architectures validate the effectiveness of our data generation pipelines in enhancing CLIP's ability to perceive negation accurately. Additionally, NegationCLIP's enhanced negation awareness has practical applications across various multimodal tasks, demonstrated by performance gains in text-to-image generation and referring image segmentation.

Detailed image captioning is essential for tasks like data generation and aiding visually impaired individuals. High-quality captions require a balance between precision and recall, which remains challenging for current multimodal large language models (MLLMs). In this work, we hypothesize that this limitation stems from weakening and increasingly noisy visual attention as responses lengthen. To address this issue, we propose SPARC (Selective Progressive Attention ReCalibration), a training-free method that enhances the contribution of visual tokens during decoding. SPARC is founded on three key observations: (1) increasing the influence of all visual tokens reduces recall; thus, SPARC selectively amplifies visual tokens; (2) as captions lengthen, visual attention becomes noisier, so SPARC identifies critical visual tokens by leveraging attention differences across time steps; (3) as visual attention gradually weakens, SPARC reinforces it to preserve its influence. Our experiments, incorporating both automated and human evaluations, demonstrate that existing methods improve the precision of MLLMs at the cost of recall. In contrast, our proposed method enhances both precision and recall with minimal computational overhead.

Utilizing the complex inter-variable causal relationships within multivariate time-series provides a promising avenue toward more robust and reliable multivariate time-series anomaly detection (MTSAD) but remains an underexplored area of research. This paper proposes Causality-Aware contrastive learning for RObust multivariate Time-Series (CAROTS), a novel MTSAD pipeline that incorporates the notion of causality into contrastive learning. CAROTS employs two data augmentors to obtain causality-preserving and -disturbing samples that serve as a wide range of normal variations and synthetic anomalies, respectively. With causality-preserving and -disturbing samples as positives and negatives, CAROTS performs contrastive learning to train an encoder whose latent space separates normal and abnormal samples based on causality. Moreover, CAROTS introduces a similarity-filtered one-class contrastive loss that encourages the contrastive learning process to gradually incorporate more semantically diverse samples with common causal relationships. Extensive experiments on five real-world and two synthetic datasets validate that the integration of causal relationships endows CAROTS with improved MTSAD capabilities. The code is available at https://github.com/kimanki/CAROTS.

Multimodal large language models (MLLMs) excel at generating highly detailed captions but often produce hallucinations. Our analysis reveals that existing hallucination detection methods struggle with detailed captions. We attribute this to the increasing reliance of MLLMs on their generated text, rather than the input image, as the sequence length grows. To address this issue, we propose a multiagent approach that leverages LLM-MLLM collaboration to correct given captions. Additionally, we introduce an evaluation framework and a benchmark dataset to facilitate the systematic analysis of detailed captions. Our experiments demonstrate that the proposed evaluation method aligns better with human judgments of factuality than existing metrics. Moreover, we show that current approaches for enhancing MLLM factuality often fail in hyper-detailed image captioning tasks. In contrast, our approach significantly enhances the factual accuracy of captions, even improving those generated by GPT-4V. Finally, we highlight a limitation of VQA-centric benchmarking by demonstrating that an MLLM’s performance on VQA benchmarks may not correlate with its ability to generate detailed image captions.

Sangwon Yu, Jongyoon Song, Bongkyu Hwang, Hoyoung Kang, Sooah Cho, Junhwa Choi, Seongho Joe, Taehee Lee, Youngjune Gwon, Sungroh Yoon. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

While large transformer models excel in predictive performance, their lack of interpretability restricts their usefulness in high-stakes domains. To remedy this, we propose the Generalized Induction-Head Model (GIM), an interpretable model for next-token prediction inspired by the observation of “induction heads” in LLMs. GIM is a retrieval-based module that identifies similar sequences in the input context by combining exact n-gram matching and fuzzy matching based on a neural similarity metric. We evaluate GIM in two settings: language modeling and fMRI response prediction. In language modeling, GIM improves next-token prediction by up to 25%p over interpretable baselines, significantly narrowing the gap with black-box LLMs. In an fMRI setting, GIM improves neural response prediction by 20% and offers insights into the language selectivity of the brain. GIM represents a significant step toward uniting interpretability and performance across domains. The code is available at https://github.com/ejkim47/generalized-induction-head.

Recent multi-modal large language models (MLLMs) often struggle to generate personalized image captions, even when trained on high-quality captions. In this work, we observe that such limitations persist in existing post-training-based MLLM personalization methods. Specifically, despite being post-tuned with large-scale caption data through supervised fine-tuning (SFT), these models frequently fail to produce faithful descriptions in real-world scenarios, such as multi-concept image captioning. However, acquiring large-scale, high-quality captions for such complex settings is both costly and difficult. To address the data-centric nature of SFT, we propose a reinforcement learning (RL)-based post-training framework. To the best of our knowledge, this is the first RL-based approach to post-train MLLMs for personalized image captioning. Our method significantly enhances both visual recognition and personalized generation capabilities of MLLMs, and consistently outperforms existing SFT-based baselines, especially in the challenging multi-concept image captioning task. Project page: https://github.com/oyt9306/RePIC

Recent studies have validated that pruning hard-to-learn examples from training improves the generalization performance of neural networks (NNs). In this study, we investigate this intriguing phenomenon---the negative effect of hard examples on generalization---in adversarial training. Particularly, we theoretically demonstrate that the increase in the difficulty of hard examples in adversarial training is significantly greater than the increase in the difficulty of easy examples. Furthermore, we verify that hard examples are only fitted through memorization of the label in adversarial training. We conduct both theoretical and empirical analyses of this memorization phenomenon, showing that pruning hard examples in adversarial training can enhance the model's robustness. However, the challenge remains in finding the optimal threshold for removing hard examples that degrade robustness performance. Based upon these observations, we propose a new approach, difficulty proportional label smoothing (DPLS), to adaptively mitigate the negative effect of hard examples, thereby improving the adversarial robustness of NNs. Notably, our experimental result indicates that our method can successfully leverage hard examples while circumventing the negative effect.

In this paper, we primarily address the issue of dialogue-form context query within the interactive text-to-image retrieval task.Our methodology, PlugIR, actively utilizes the general instruction-following capability of LLMs in two ways.First, by reformulating the dialogueform context, we eliminate the necessity of fine-tuning a retrieval model on existing visual dialogue data, thereby enabling the use of any arbitrary black-box model.Second, we construct the LLM questioner to generate nonredundant questions about the attributes of the target image, based on the information of retrieval candidate images in the current context.This approach mitigates the issues of noisiness and redundancy in the generated questions.Beyond our methodology, we propose a novel evaluation metric, Best log Rank Integral (BRI), for a comprehensive assessment of the interactive retrieval system.PlugIR demonstrates superior performance compared to both zero-shot and fine-tuned baselines in various benchmarks.Additionally, the two methodologies comprising PlugIR can be flexibly applied together or separately in various situations.Our codes are available at https://github.com/ Saehyung-Lee/PlugIR.

Controlled text generation is very important for the practical use of language models because it ensures that the produced text includes only the desired attributes from a specific domain or dataset.Existing methods, however, are inapplicable to black-box models or suffer a significant trade-off between controlling the generated text and maintaining its fluency.This paper introduces the Score-based Progressive Editor (ScoPE), a novel approach designed to overcome these issues.ScoPE modifies the context at the token level during the generation process of a backbone language model.This modification guides the subsequent text to naturally include the target attributes.To facilitate this process, ScoPE employs a training objective that maximizes a target score, thoroughly considering both the ability to guide the text and its fluency.Experimental results on diverse controlled generation tasks demonstrate that ScoPE can effectively regulate the attributes of the generated text while fully utilizing the capability of the backbone large language models.Our codes are available at https://github.com/ysw1021/ScoPE.

In this paper, we address the problem of unsupervised video anomaly detection (UVAD). The task aims to detect abnormal events in test video using unlabeled videos as training data. The presence of anomalies in the training data poses a significant challenge in this task, particularly because they form clusters in the feature space. We refer to this property as the "Anomaly Cluster" issue. The condensed nature of these anomalies makes it difficult to distinguish between normal and abnormal data in the training set. Consequently, training conventional anomaly detection techniques using an unlabeled dataset often leads to sub-optimal results. To tackle this difficulty, we propose a new method called Cleansed k-Nearest Neighbor (CKNN), which explicitly filters out the Anomaly Clusters by cleansing the training dataset. Following the k-nearest neighbor algorithm in the feature space provides powerful anomaly detection capability. Although the identified Anomaly Cluster issue presents a significant challenge to applying k-nearest neighbor in UVAD, our proposed cleansing scheme effectively addresses this problem. We evaluate the proposed method on various benchmark datasets and demonstrate that CKNN outperforms the previous state-of-the-art UVAD method by up to 8.5% (from 82.0 to 89.0) in terms of AUROC. Moreover, we emphasize that the performance of the proposed method is comparable to that of the state-of-the-art method trained using anomaly-free data.

Jongyoon Song, Nohil Park, Bongkyu Hwang, Jaewoong Yun, Seongho Joe, Youngjune Gwon, Sungroh Yoon. Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Test-time adaptation (TTA) fine-tunes pre-trained deep neural networks for unseen test data. The primary challenge of TTA is limited access to the entire test dataset during online updates, causing error accumulation. To mitigate it, TTA methods have utilized the model output's entropy as a confidence metric that aims to determine which samples have a lower likelihood of causing error. Through experimental studies, however, we observed the unreliability of entropy as a confidence metric for TTA under biased scenarios and theoretically revealed that it stems from the neglect of the influence of latent disentangled factors of data on predictions. Building upon these findings, we introduce a novel TTA method named Destroy Your Object (DeYO), which leverages a newly proposed confidence metric named Pseudo-Label Probability Difference (PLPD). PLPD quantifies the influence of the shape of an object on prediction by measuring the difference between predictions before and after applying an object-destructive transformation. DeYO consists of sample selection and sample weighting, which employ entropy and PLPD concurrently. For robust adaptation, DeYO prioritizes samples that dominantly incorporate shape information when making predictions. Our extensive experiments demonstrate the consistent superiority of DeYO over baseline methods across various scenarios, including biased and wild. Project page is publicly available at https://whitesnowdrop.github.io/DeYO/.

In the face of the deep learning model's vulnerability to domain shift, source-free domain adaptation (SFDA) methods have been proposed to adapt models to new, unseen target domains without requiring access to source domain data. Although the potential benefits of applying data augmentation to SFDA are attractive, several challenges arise such as the dependence on prior knowledge of class-preserving transformations and the increase in memory and computational requirements. In this paper, we propose Source-free Domain Adaptation Through the Lens of Data Augmentation (SF(DA)$^2$), a novel approach that leverages the benefits of data augmentation without suffering from these challenges. We construct an augmentation graph in the feature space of the pretrained model using the neighbor relationships between target features and propose spectral neighborhood clustering to identify partitions in the prediction space. Furthermore, we propose implicit feature augmentation and feature disentanglement as regularization loss functions that effectively utilize class semantic information within the feature space. These regularizers simulate the inclusion of an unlimited number of augmented target features into the augmentation graph while minimizing computational and memory demands. Our method shows superior adaptation performance in SFDA scenarios, including 2D image and 3D point cloud datasets and a highly imbalanced dataset.

The disparity in accuracy between classes in standard training is amplified during adversarial training, a phenomenon termed the robust fairness problem. Existing methodologies aimed to enhance robust fairness by sacrificing the model's performance on easier classes in order to improve its performance on harder ones. However, we observe that under adversarial attacks, the majority of the model's predictions for samples from the worst class are biased towards classes similar to the worst class, rather than towards the easy classes. Through theoretical and empirical analysis, we demonstrate that robust fairness deteriorates as the distance between classes decreases. Motivated by these insights, we introduce the Distance-Aware Fair Adversarial Training (DAFA) methodology, which addresses robust fairness by taking into account the similarities between classes. Specifically, our method assigns distinct adversarial margins and loss weights to each class and adjusts them to encourage a trade-off in robustness among similar classes. Experimental results across various datasets demonstrate that our method not only maintains average robust accuracy but also significantly improves the worst robust accuracy, indicating a marked improvement in robust fairness compared to existing methods.

Sequence modeling faces challenges in capturing long-range dependencies across diverse tasks. Recent linear and transformer-based forecasters have shown superior performance in time series forecasting. However, they are constrained by their inherent inability to effectively address long-range dependencies in time series data, primarily due to using fixed-size inputs for prediction. Furthermore, they typically sacrifice essential temporal correlation among consecutive training samples by shuffling them into mini-batches. To overcome these limitations, we introduce a fast and effective Spectral Attention mechanism, which preserves temporal correlations among samples and facilitates the handling of long-range information while maintaining the base model structure. Spectral Attention preserves long-period trends through a low-pass filter and facilitates gradient to flow between samples. Spectral Attention can be seamlessly integrated into most sequence models, allowing models with fixed-sized look-back windows to capture long-range dependencies over thousands of steps. Through extensive experiments on 11 real-world time series datasets using 7 recent forecasting models, we consistently demonstrate the efficacy of our Spectral Attention mechanism, achieving state-of-the-art results.

Recent work shows promising results in expanding the capabilities of large language models (LLM) to directly understand and synthesize speech. However, an LLM-based strategy for modeling spoken dialogs remains elusive, calling for further investigation. This paper introduces an extensive speech-text LLM framework, the Unified Spoken Dialog Model (USDM), designed to generate coherent spoken responses with naturally occurring prosodic features relevant to the given input speech without relying on explicit automatic speech recognition (ASR) or text-to-speech (TTS) systems. We have verified the inclusion of prosody in speech tokens that predominantly contain semantic information and have used this foundation to construct a prosody-infused speech-text model. Additionally, we propose a generalized speech-text pretraining scheme that enhances the capture of cross-modal semantics. To construct USDM, we fine-tune our speech-text model on spoken dialog data using a multi-step spoken dialog template that stimulates the chain-of-reasoning capabilities exhibited by the underlying LLM. Automatic and human evaluations on the DailyTalk dataset demonstrate that our approach effectively generates natural-sounding spoken responses, surpassing previous and cascaded baselines. Our code and checkpoints are available at https://github.com/naver-ai/usdm.

In our study, we explore methods for detecting unwanted content lurking in visual datasets. We provide a theoretical analysis demonstrating that a model capable of successfully partitioning visual data can be obtained using only textual data. Based on the analysis, we propose Hassle-Free Textual Training (HFTT), a streamlined method capable of acquiring detectors for unwanted visual content, using only textual data in conjunction with pre-trained vision-language models. HFTT features an innovative objective function that significantly reduces the necessity for human involvement in data annotation. Furthermore, HFTT employs a clever textual data synthesis method, effectively emulating the integration of unknown visual data distribution into the training process at no extra cost. The unique characteristics of HFTT extend its utility beyond traditional out-of-distribution detection, making it applicable to tasks that address more abstract concepts. We complement our analyses with experiments in hateful image detection and out-of-distribution detection. Our codes are available at https://github.com/HFTT-anonymous/HFTT.

Estimating the homography between two images is crucial for mid- or high-level vision tasks, such as image stitching and fusion. However, using supervised learning methods is often challenging or costly due to the difficulty of collecting ground-truth data. In response, unsupervised learning approaches have emerged. Most early methods, though, assume that the given image pairs are from the same camera or have minor lighting differences. Consequently, while these methods perform effectively under such conditions, they generally fail when input image pairs come from different domains, referred to as multimodal image pairs. To address these limitations, we propose AltO, an unsupervised learning framework for estimating homography in multimodal image pairs. Our method employs a two-phase alternating optimization framework, similar to Expectation-Maximization (EM), where one phase reduces the geometry gap and the other addresses the modality gap. To handle these gaps, we use Barlow Twins loss for the modality gap and propose an extended version, Geometry Barlow Twins, for the geometry gap. As a result, we demonstrate that our method, AltO, can be trained on multimodal datasets without any ground-truth data. It not only outperforms other unsupervised methods but is also compatible with various architectures of homography estimators. The source code can be found at: https://github.com/songsang7/AltO

Imitation learning, in which learning is performed by demonstration, has been studied and advanced for sequential decision-making tasks in which a reward function is not predefined. However, imitation learning methods still require numerous expert demonstration samples to successfully imitate an expert's behavior. To improve sample efficiency, we utilize self-supervised representation learning, which can generate vast training signals from the given data. In this study, we propose a self-supervised representation-based adversarial imitation learning method to learn state and action representations that are robust to diverse distortions and temporally predictive, on non-image control tasks. In particular, in comparison with existing self-supervised learning methods for tabular data, we propose a different corruption method for state and action representations that is robust to diverse distortions. We theoretically and empirically observe that making an informative feature manifold with less sample complexity significantly improves the performance of imitation learning. The proposed method shows a 39% relative improvement over existing adversarial imitation learning methods on MuJoCo in a setting limited to 100 expert state-action pairs. Moreover, we conduct comprehensive ablations and additional experiments using demonstrations with varying optimality to provide insights into a range of factors.

Obtaining accurate pixel-level localization from class labels is a crucial process in weakly supervised semantic segmentation and object localization. Attribution maps from a trained classifier are widely used to provide pixel-level localization, but their focus tends to be restricted to a small discriminative region of the target object. An AdvCAM is an attribution map of an image that is manipulated to increase the classification score produced by a classifier before the final softmax or sigmoid layer. This manipulation is realized in an anti-adversarial manner, so that the original image is perturbed along pixel gradients in directions opposite to those used in an adversarial attack. This process enhances non-discriminative yet class-relevant features, which make an insufficient contribution to previous attribution maps, so that the resulting AdvCAM identifies more regions of the target object. In addition, we introduce a new regularization procedure that inhibits the incorrect attribution of regions unrelated to the target object and the excessive concentration of attributions on a small region of the target object. Our method achieves a new state-of-the-art performance in weakly and semi-supervised semantic segmentation, on both the PASCAL VOC 2012 and MS COCO 2014 datasets. In weakly supervised object localization, it achieves a new state-of-the-art performance on the CUB-200-2011 and ImageNet-1K datasets.

Jointly fine-tuning a Pre-trained Language Model (PLM) on a pre-defined set of tasks with in-context instructions has been proven to improve its generalization performance, allowing us to build a universal language model that can be deployed across task boundaries.In this work, we explore for the first time whether this attractive property of in-context instruction learning can be extended to a scenario in which tasks are fed to the target PLM in a sequential manner.The primary objective of so-called lifelong in-context instruction learning is to improve the target PLM's instance-and task-level generalization performance as it observes more tasks.DYNAINST, the proposed method to lifelong in-context instruction learning, achieves noticeable improvements in both types of generalization, nearly reaching the upper bound performance obtained through joint training.

Several recent studies have elucidated why knowledge distillation (KD) improves model performance. However, few have researched the other advantages of KD in addition to its improving model performance. In this study, we have attempted to show that KD enhances the interpretability as well as the accuracy of models. We measured the number of concept detectors identified in network dissection for a quantitative comparison of model interpretability. We attributed the improvement in interpretability to the class-similarity information transferred from the teacher to student models. First, we confirmed the transfer of class-similarity information from the teacher to student model via logit distillation. Then, we analyzed how class-similarity information affects model interpretability in terms of its presence or absence and degree of similarity information. We conducted various quantitative and qualitative experiments and examined the results on different datasets, different KD methods, and according to different measures of interpretability. Our research showed that KD models by large models could be used more reliably in various fields. The code is available at https://github.com/Rok07/KD_XAI.git.

Interpretable models are designed to make decisions in a human-interpretable manner. Representatively, Concept Bottleneck Models (CBM) follow a two-step process of concept prediction and class prediction based on the predicted concepts. CBM provides explanations with high-level concepts derived from concept predictions; thus, reliable concept predictions are important for trustworthiness. In this study, we address the ambiguity issue that can harm reliability. While the existence of a concept can often be ambiguous in the data, CBM predicts concepts deterministically without considering this ambiguity. To provide a reliable interpretation against this ambiguity, we propose Probabilistic Concept Bottleneck Models (ProbCBM). By leveraging probabilistic concept embeddings, ProbCBM models uncertainty in concept prediction and provides explanations based on the concept and its corresponding uncertainty. This uncertainty enhances the reliability of the explanations. Furthermore, as class uncertainty is derived from concept uncertainty in ProbCBM, we can explain class uncertainty by means of concept uncertainty. Code is publicly available at https://github.com/ejkim47/prob-cbm.

Visual Prompt Tuning (VPT) is an effective tuning method for adapting pretrained Vision Transformers (ViTs) to downstream tasks. It leverages extra learnable tokens, known as prompts, which steer the frozen pretrained ViTs. Although VPT has demonstrated its applicability with supervised vision transformers, it often underperforms with self-supervised ones. Through empirical observations, we deduce that the effectiveness of VPT hinges largely on the ViT blocks with which the prompt tokens interact. Specifically, VPT shows improved performance on image classification tasks for MAE and MoCo v3 when the prompt tokens are inserted into later blocks rather than the first block. These observations suggest that there exists an optimal location of blocks for the insertion of prompt tokens. Unfortunately, identifying the optimal blocks for prompts within each self-supervised ViT for diverse future scenarios is a costly process. To mitigate this problem, we propose a simple yet effective method that learns a gate for each ViT block to adjust its intervention into the prompt tokens. With our method, prompt tokens are selectively influenced by blocks that require steering for task adaptation. Our method outperforms VPT variants in FGVC and VTAB image classification and ADE20K semantic segmentation. The code is available at https://github.com/ryongithub/GatedPromptTuning.

Although supervised image denoising networks have shown remarkable performance on synthesized noisy images, they often fail in practice due to the difference between real and synthesized noise. Since clean-noisy image pairs from the real world are extremely costly to gather, self-supervised learning, which utilizes noisy input itself as a target, has been studied. To prevent a self-supervised denoising model from learning identical mapping, each output pixel should not be influenced by its corresponding input pixel; This requirement is known as J-invariance. Blind-spot networks (BSNs) have been a prevalent choice to ensure J-invariance in self-supervised image denoising. However, constructing variations of BSNs by injecting additional operations such as downsampling can expose blinded information, thereby violating J-invariance. Consequently, convolutions designed specifically for BSNs have been allowed only, limiting architectural flexibility. To overcome this limitation, we propose PUCA, a novel J-invariant U-Net architecture, for self-supervised denoising. PUCA leverages patch-unshuffle/shuffle to dramatically expand receptive fields while maintaining J-invariance and dilated attention blocks (DABs) for global context incorporation. Experimental results demonstrate that PUCA achieves state-of-the-art performance, outperforming existing methods in self-supervised image denoising.

Continual learning (CL) aims to incrementally learn multiple tasks that are presented sequentially. The significance of CL lies not only in the practical importance but also in studying the learning mechanisms of humans who are excellent continual learners. While most research on CL has been done on structured data such as images, there is a lack of research on CL for abstract logical concepts such as counting, sorting, and arithmetic, which humans learn gradually over time in the real world. In this work, for the first time, we introduce novel algorithmic reasoning (AR) methodology for continual tasks of abstract concepts: CLeAR. Our methodology proposes a one-to-many mapping of input distribution to a shared mapping space, which allows the alignment of various tasks of different dimensions and shared semantics. Our tasks of abstract logical concepts, in the form of formal language, can be classified into Chomsky hierarchies based on their difficulty. In this study, we conducted extensive experiments consisting of 15 tasks with various levels of Chomsky hierarchy, ranging from in-hierarchy to inter-hierarchy scenarios. CLeAR not only achieved near zero forgetting but also improved accuracy during following tasks, a phenomenon known as backward transfer, while previous CL methods designed for image classification drastically failed.

While recent large-scale neural codec language models have shown significant improvement in zero-shot TTS by training on thousands of hours of data, they suffer from drawbacks such as a lack of robustness, slow sampling speed similar to previous autoregressive TTS methods, and reliance on pre-trained neural codec representations. Our work proposes P-Flow, a fast and data-efficient zero-shot TTS model that uses speech prompts for speaker adaptation. P-Flow comprises a speech-prompted text encoder for speaker adaptation and a flow matching generative decoder for high-quality and fast speech synthesis. Our speech-prompted text encoder uses speech prompts and text input to generate speaker-conditional text representation. The flow matching generative decoder uses the speaker-conditional output to synthesize high-quality personalized speech significantly faster than in real-time. Unlike the neural codec language models, we specifically train P-Flow on LibriTTS dataset using a continuous mel-representation. Through our training method using continuous speech prompts, P-Flow matches the speaker similarity performance of the large-scale zero-shot TTS models with two orders of magnitude less training data and has more than 20$\times$ faster sampling speed. Our results show that P-Flow has better pronunciation and is preferred in human likeness and speaker similarity to its recent state-of-the-art counterparts, thus defining P-Flow as an attractive and desirable alternative. We provide audio samples on our demo page: [https://research.nvidia.com/labs/adlr/projects/pflow](https://research.nvidia.com/labs/adlr/projects/pflow)

The rapid advancement and widespread use of large language models (LLMs) have raised significant concerns regarding the potential leakage of personally identifiable information (PII). These models are often trained on vast quantities of web-collected data, which may inadvertently include sensitive personal data. This paper presents ProPILE, a novel probing tool designed to empower data subjects, or the owners of the PII, with awareness of potential PII leakage in LLM-based services. ProPILE lets data subjects formulate prompts based on their own PII to evaluate the level of privacy intrusion in LLMs. We demonstrate its application on the OPT-1.3B model trained on the publicly available Pile dataset. We show how hypothetical data subjects may assess the likelihood of their PII being included in the Pile dataset being revealed. ProPILE can also be leveraged by LLM service providers to effectively evaluate their own levels of PII leakage with more powerful prompts specifically tuned for their in-house models. This tool represents a pioneering step towards empowering the data subjects for their awareness and control over their own data on the web.

Successful detection of Out-of-Distribution (OoD) data is becoming increasingly important to ensure safe deployment of neural networks. One of the main challenges in OoD detection is that neural networks output overconfident predictions on OoD data, make it difficult to determine OoD-ness of data solely based on their predictions. Outlier exposure addresses this issue by introducing an additional loss that encourages low-confidence predictions on OoD data during training. While outlier exposure has shown promising potential in improving OoD detection performance, all previous studies on outlier exposure have been limited to utilizing visual outliers. Drawing inspiration from the recent advancements in vision-language pre-training, this paper venture out to the uncharted territory of textual outlier exposure. First, we uncover the benefits of using textual outliers by replacing real or virtual outliers in the image-domain with textual equivalents. Then, we propose various ways of generating preferable textual outliers. Our extensive experiments demonstrate that generated textual outliers achieve competitive performance on large-scale OoD and hard OoD benchmarks. Furthermore, we conduct empirical analyses of textual outliers to provide primary criteria for designing advantageous textual outliers: near-distribution, descriptiveness, and inclusion of visual semantics.

Amyloid positivity is an early indicator of Alzheimer's disease and is necessary to determine the disease. In this study, a deep generative model is utilized to predict the amyloid positivity of cognitively normal individuals using proxy measures, such as structural MRI scans, demographic variables, and cognitive scores, instead of invasive direct measurements. Through its remarkable efficacy in handling imperfect datasets caused by missing data or labels, and imbalanced classes, the model outperforms previous studies and widely used machine learning approaches with an AUROC of 0.8609. Furthermore, this study illuminates the model's adaptability to diverse clinical scenarios, even when feature sets or diagnostic criteria differ from the training data. We identify the brain regions and variables that contribute most to classification, including the lateral occipital lobes, posterior temporal lobe, and APOE ϵ4 allele. Taking advantage of deep generative models, our approach can not only provide inexpensive, non-invasive, and accurate diagnostics for preclinical Alzheimer's disease, but also meet real-world requirements for clinical translation of a deep learning model, including transferability and interpretability.

Recently, the field of steganography has experienced rapid developments based on deep learning (DL). DL based steganography distributes secret information over all the available bits of the cover image, thereby posing difficulties in using conventional steganalysis methods to detect, extract or remove hidden secret images. However, our proposed framework is the first to effectively disable covert communications and transactions that use DL based steganography. We propose a DL based steganalysis technique that effectively removes secret images by restoring the distribution of the original images. We formulate a problem and address it by exploiting sophisticated pixel distributions and an edge distribution of images by using a deep neural network. Based on the given information, we remove the hidden secret information at the pixel level. We evaluate our technique by comparing it with conventional steganalysis methods using three public benchmarks. As the decoding method of DL based steganography is approximate (lossy) and is different from the decoding method of conventional steganography, we also introduce a new quantitative metric called the destruction rate (DT). The experimental results demonstrate performance improvements of 10–20$\%$in both the decoded rate and the DT.

In recent years, proposed studies on time-series anomaly detection (TAD) report high F1 scores on benchmark TAD datasets, giving the impression of clear improvements in TAD. However, most studies apply a peculiar evaluation protocol called point adjustment (PA) before scoring. In this paper, we theoretically and experimentally reveal that the PA protocol has a great possibility of overestimating the detection performance; even a random anomaly score can easily turn into a state-of-the-art TAD method. Therefore, the comparison of TAD methods after applying the PA protocol can lead to misguided rankings. Furthermore, we question the potential of existing TAD methods by showing that an untrained model obtains comparable detection performance to the existing methods even when PA is forbidden. Based on our findings, we propose a new baseline and an evaluation protocol. We expect that our study will help a rigorous evaluation of TAD and lead to further improvement in future researches.

Sangwon Yu, Jongyoon Song, Heeseung Kim, Seongmin Lee, Woo-Jong Ryu, Sungroh Yoon. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Membership inference (MI) determines whether a given data point is involved in the training of target machine learning model. Thus, the notion of MI relies on both the data feature and the model. The existing MI methods focus on the model only. We introduce a membership feature disentanglement network (MFDN) to approach MI from the perspective of data features. We assume that the data features can be disentangled into the membership features and class features. The membership features are those that enable MI, and class features refer to those that the network is trying to learn. MFDN disentangles these features by adversarial games between the encoders and auxiliary critic networks. It also visualizes the membership features using an inductive bias from the perspective of MI. We perform empirical evaluations to demonstrate that MFDN can disentangle membership features and class features.

Making deep neural networks available as a service introduces privacy problems, for which homomorphic encryption of both model and user data potentially offers the solution at the highest privacy level. However, the difficulty of operating on homomorphically encrypted data has hitherto limited the range of operations available and the depth of networks. We introduce an extended CKKS scheme MatHEAAN to provide efficient matrix representations and operations together with improved noise control. Using the MatHEAAN we developed a deep sequential model with a gated recurrent unit called MatHEGRU. We evaluated the proposed model using sequence modeling, regression, and classification of images and genome sequences. We show that the hidden states of the encrypted model, as well as the results, are consistent with a plaintext model.

Diffusion models learn to restore noisy data, which is corrupted with different levels of noise, by optimizing the weighted sum of the corresponding loss terms, i.e., denoising score matching loss. In this paper, we show that restoring data corrupted with certain noise levels offers a proper pretext task for the model to learn rich visual concepts. We propose to prioritize such noise levels over other levels during training, by redesigning the weighting scheme of the objective function. We show that our simple redesign of the weighting scheme significantly improves the performance of diffusion models regardless of the datasets, architectures, and sampling strategies.

Weakly supervised object localization aims to find a target object region in a given image with only weak supervision, such as image-level labels. Most existing methods use a class activation map (CAM) to generate a localization map; however, a CAM identifies only the most discriminative parts of a target object rather than the entire object region. In this work, we find the gap between classification and localization in terms of the misalignment of the directions between an input feature and a class-specific weight. We demonstrate that the misalignment suppresses the activation of CAM in areas that are less discriminative but belong to the target object. To bridge the gap, we propose a method to align feature directions with a class-specific weight. The proposed method achieves a state-of-the-art localization performance on the CUB-200-2011 and ImageNet-1K benchmarks.

The images captured by under-display cameras (UDCs) are degraded by the screen in front of them. We model this degradation in terms of a) diffraction by the pixel grid, which attenuates high-spatial-frequency components of the image; and b) diffuse intensity and color changes caused by the multiple thin-film layers in an OLED, which modulate the low-spatial-frequency components of the image. We introduce a deep neural network with two branches to reverse each type of degradation, which is more effective than performing both restorations in a single forward network. We also propose an affine transform connection to replace the skip connection used in most existing DNNs for restoring UDC images. Confining the solution space to the linear transform domain reduces the blurring caused by convolution; and any gross color shift in the training images is eliminated by inverse color filtering. Trained on three datasets of UDC images, our network outperformed existing methods in terms of measures of distortion and of perceived image quality.

Weakly supervised semantic segmentation (WSSS) methods are often built on pixel-level localization maps obtained from a classifier. However, training on class labels only, classifiers suffer from the spurious correlation between fore-ground and background cues (e.g. train and rail), fundamentally bounding the performance of WSSS. There have been previous endeavors to address this issue with additional supervision. We propose a novel source of information to distinguish foreground from the background: Out-of-Distribution (OoD) data, or images devoid of foreground object classes. In particular, we utilize the hard OoDs that the classifier is likely to make false-positive predictions. These samples typically carry key visual features on the background (e.g. rail) that the classifiers often confuse as foreground (e.g. train), so these cues let classifiers correctly suppress spurious background cues. Acquiring such hard OoDs does not require an extensive amount of annotation efforts; it only incurs a few additional image-level labeling costs on top of the original efforts to collect class labels. We propose a method, W-OoD, for utilizing the hard OoDs. W-OoD achieves state-of-the-art performance on Pascal VOC 2012. The code is available at: https://github.com/naver-ai/w-ood.

In Neural Architecture Search (NAS), reducing the cost of architecture evaluation remains one of the most crucial challenges. Among a plethora of efforts to bypass training of each candidate architecture to convergence for evaluation, the Neural Tangent Kernel (NTK) is emerging as a promising theoretical framework that can be utilized to estimate the performance of a neural architecture at initialization. In this work, we revisit several at-initialization metrics that can be derived from the NTK and reveal their key short-comings. Then, through the empirical analysis of the time evolution of NTK, we deduce that modern neural architectures exhibit highly non-linear characteristics, making the NTK-based metrics incapable of reliably estimating the performance of an architecture without some amount of training. To take such non-linear characteristics into account, we introduce Label-Gradient Alignment (LGA), a novel NTK-based metric whose inherent formulation allows it to capture the large amount of non-linear advantage present in modern neural architectures. With minimal amount of training, LGA obtains a meaningful level of rank correlation with the final test accuracy of an architecture. Lastly, we demonstrate that LGA, complemented with few epochs of training, successfully guides existing search algorithms to achieve competitive search performances with significantly less search cost. The code is available at: https://github.com/nute11amok/DemystifyingNTK.

Source-free unsupervised domain adaptation (SFUDA) aims to obtain high performance in the unlabeled target domain using the pre-trained source model, not the source data. Existing SFUDA methods assign the same importance to all target samples, which is vulnerable to incorrect pseudo-labels. To differentiate between sample importance, in this study, we propose a novel sample-wise confidence score, the Joint Model-Data Structure (JMDS) score for SFUDA. Unlike existing confidence scores that use only one of the source or target domain knowledge, the JMDS score uses both knowledge. We then propose a Confidence score Weighting Adaptation using the JMDS (CoWA-JMDS) framework for SFUDA. CoWA-JMDS consists of the JMDS scores as sample weights and weight Mixup that is our proposed variant of Mixup. Weight Mixup promotes the model make more use of the target domain knowledge. The experimental results show that the JMDS score outperforms the existing confidence scores. Moreover, CoWA-JMDS achieves state-of-the-art performance on various SFUDA scenarios: closed, open, and partial-set scenarios.

We propose Guided-TTS, a high-quality text-to-speech (TTS) model that does not require any transcript of target speaker using classifier guidance. Guided-TTS combines an unconditional diffusion probabilistic model with a separately trained phoneme classifier for classifier guidance. Our unconditional diffusion model learns to generate speech without any context from untranscribed speech data. For TTS synthesis, we guide the generative process of the diffusion model with a phoneme classifier trained on a large-scale speech recognition dataset. We present a norm-based scaling method that reduces the pronunciation errors of classifier guidance in Guided-TTS. We show that Guided-TTS achieves a performance comparable to that of the state-of-the-art TTS model, Grad-TTS, without any transcript for LJSpeech. We further demonstrate that Guided-TTS performs well on diverse datasets including a long-form untranscribed dataset.

Recent studies have demonstrated that gradient matching-based dataset synthesis, or dataset condensation (DC), methods can achieve state-of-theart performance when applied to data-efficient learning tasks. However, in this study, we prove that the existing DC methods can perform worse than the random selection method when taskirrelevant information forms a significant part of the training dataset. We attribute this to the lack of participation of the contrastive signals between the classes resulting from the class-wise gradient matching strategy. To address this problem, we propose Dataset Condensation with Contrastive signals (DCC) by modifying the loss function to enable the DC methods to effectively capture the differences between classes. In addition, we analyze the new loss function in terms of training dynamics by tracking the kernel velocity. Furthermore, we introduce a bi-level warm-up strategy to stabilize the optimization. Our experimental results indicate that while the existing methods are ineffective for fine-grained image classification tasks, the proposed method can successfully generate informative synthetic datasets for the same tasks. Moreover, we demonstrate that the proposed method outperforms the baselines even on benchmark datasets such as SVHN, CIFAR-10, and CIFAR-100. Finally, we demonstrate the high applicability of the proposed method by applying it to continual learning tasks.

Spiking neural networks (SNNs) that mimic information transmission in the brain can energy-efficiently process spatio-temporal information through discrete and sparse spikes, thereby receiving considerable attention. To improve accuracy and energy efficiency of SNNs, most previous studies have focused solely on training methods, and the effect of architecture has rarely been studied. We investigate the design choices used in the previous studies in terms of the accuracy and number of spikes and figure out that they are not best-suited for SNNs. To further improve the accuracy and reduce the spikes generated by SNNs, we propose a spike-aware neural architecture search framework called AutoSNN. We define a search space consisting of architectures without undesirable design choices. To enable the spike-aware architecture search, we introduce a fitness that considers both the accuracy and number of spikes. AutoSNN successfully searches for SNN architectures that outperform hand-crafted SNNs in accuracy and energy efficiency. We thoroughly demonstrate the effectiveness of AutoSNN on various datasets including neuromorphic datasets.

Personalized federated learning is aimed at allowing numerous clients to train personalized models while participating in collaborative training in a communication-efficient manner without exchanging private data. However, many personalized federated learning algorithms assume that clients have the same neural network architecture, and those for heterogeneous models remain understudied. In this study, we propose a novel personalized federated learning method called federated classifier averaging (FedClassAvg). Deep neural networks for supervised learning tasks consist of feature extractor and classifier layers. FedClassAvg aggregates classifier weights as an agreement on decision boundaries on feature spaces so that clients with not independently and identically distributed (non-iid) data can learn about scarce labels. In addition, local feature representation learning is applied to stabilize the decision boundaries and improve the local feature extraction capabilities for clients. While the existing methods require the collection of auxiliary data or model weights to generate a counterpart, FedClassAvg only requires clients to communicate with a couple of fully connected layers, which is highly communication-efficient. Moreover, FedClassAvg does not require extra optimization problems such as knowledge transfer, which requires intensive computation overhead. We evaluated FedClassAvg through extensive experiments and demonstrated it outperforms the current state-of-the-art algorithms on heterogeneous personalized federated learning tasks.

MOTIVATION: MicroRNAs (miRNAs) play pivotal roles in gene expression regulation by binding to target sites of messenger RNAs (mRNAs). While identifying functional targets of miRNAs is of utmost importance, their prediction remains a great challenge. Previous computational algorithms have major limitations. They use conservative candidate target site (CTS) selection criteria mainly focusing on canonical site types, rely on laborious and time-consuming manual feature extraction, and do not fully capitalize on the information underlying miRNA-CTS interactions. RESULTS: In this article, we introduce TargetNet, a novel deep learning-based algorithm for functional miRNA target prediction. To address the limitations of previous approaches, TargetNet has three key components: (i) relaxed CTS selection criteria accommodating irregularities in the seed region, (ii) a novel miRNA-CTS sequence encoding scheme incorporating extended seed region alignments and (iii) a deep residual network-based prediction model. The proposed model was trained with miRNA-CTS pair datasets and evaluated with miRNA-mRNA pair datasets. TargetNet advances the previous state-of-the-art algorithms used in functional miRNA target classification. Furthermore, it demonstrates great potential for distinguishing high-functional miRNA targets. AVAILABILITY AND IMPLEMENTATION: The codes and pre-trained models are available at https://github.com/mswzeus/TargetNet.

Recent advances in next-generation sequencing technologies have led to the successful insertion of video information into DNA using synthesized oligonucleotides. Several attempts have been made to embed larger data into living organisms. This process of embedding messages is called steganography and it is used for hiding and watermarking data to protect intellectual property. In contrast, steganalysis is a group of algorithms that serves to detect hidden information from covert media. Various methods have been developed to detect messages embedded in conventional covert channels. However, conventional steganalysis algorithms are mostly limited to common covert media. Most common detection approaches, such as frequency analysis-based methods, often overlook important signals when directly applied to DNA steganography and are easily bypassed by recently developed steganography techniques. To address the limitations of conventional approaches, a sequence-learning-based malicious DNA sequence analysis method based on neural networks has been proposed. The proposed method learns intrinsic distributions and identifies distribution variations using a classification score to predict whether a sequence is to be a coding or non-coding sequence. Based on our experiments and results, we have developed a framework to safeguard security against DNA steganography.

Learning classifiers with imbalanced data can be strongly biased toward the majority class. To address this issue, several methods have been proposed using generative adversarial networks (GANs). Existing GAN-based methods, however, do not effectively utilize the relationship between a classifier and a generator. This article proposes a novel three-player structure consisting of a discriminator, a generator, and a classifier, along with decision boundary regularization. Our method is distinctive in which the generator is trained in cooperation with the classifier to provide minority samples that gradually expand the minority decision region, improving performance for imbalanced data classification. The proposed method outperforms the existing methods on real data sets as well as synthetic imbalanced data sets.

Automated diagnosis using deep neural networks in chest radiography can help radiologists detect life-threatening diseases. However, existing methods only provide predictions without accurate explanations, undermining the trustworthiness of the diagnostic methods. Here, we present XProtoNet, a globally and locally interpretable diagnosis framework for chest radiography. XProtoNet learns representative patterns of each disease from X-ray images, which are prototypes, and makes a diagnosis on a given X-ray image based on the patterns. It predicts the area where a sign of the disease is likely to appear and compares the features in the predicted area with the prototypes. It can provide a global explanation, the prototype, and a local explanation, how the prototype contributes to the prediction of a single image. Despite the constraint for interpretability, XProtoNet achieves state-of-the-art classification performance on the public NIH chest X-ray dataset.

Weakly supervised semantic segmentation produces a pixel-level localization from class labels; but a classifier trained on such labels is likely to restrict its focus to a small discriminative region of the target object. AdvCAM is an attribution map of an image that is manipulated to increase the classification score produced by a classifier. This manipulation is realized in an anti-adversarial manner, which perturbs the original images along pixel gradients in the opposite direction from those used in an adversarial attack. It forces regions initially considered not to be discriminative to become involved in subsequent classifications, and produces attribution maps that successively identify more regions of the target object. In addition, we introduce a new regularization procedure that inhibits both the incorrect attribution of regions unrelated to the target object and excessive concentration of attributions on a small region of that object. Our method is a post-hoc analysis of a trained classifier, which does not need to be altered or retrained. On PASCAL VOC 2012 test images, we achieve mIoUs of 68.0 and 76.9 for weakly and semi-supervised semantic segmentation respectively, which represent a new state-of-the-art.

Weakly supervised segmentation methods using bounding box annotations focus on obtaining a pixel-level mask from each box containing an object. Existing methods typically depend on a class-agnostic mask generator, which operates on the low-level information intrinsic to an image. In this work, we utilize higher-level information from the behavior of a trained object detector, by seeking the smallest areas of the image from which the object detector produces almost the same result as it does from the whole image. These areas constitute a bounding-box attribution map (BBAM), which identifies the target object in its bounding box and thus serves as pseudo ground-truth for weakly supervised semantic and instance segmentation. This approach significantly outperforms recent comparable techniques on both the PASCAL VOC and MS COCO benchmarks in weakly supervised semantic and instance segmentation. In addition, we provide a detailed analysis of our method, offering deeper insight into the behavior of the BBAM.

Spiking neural networks (SNNs) have emerged as energy-efficient neural networks with temporal information. SNNs have shown a superior efficiency on neuromorphic devices, but the devices are susceptible to noise, which hinders them from being applied in real-world applications. Several studies have increased noise robustness, but most of them considered neither deep SNNs nor temporal information. In this paper, we investigate the effect of noise on deep SNNs with various neural coding methods and present a noise-robust deep SNN with temporal information. With the proposed methods, we have achieved a deep SNN that is efficient and robust to spike deletion and jitter.

Non-autoregressive neural machine translation (NART) models suffer from the multi-modality problem which causes translation inconsistency such as token repetition.Most recent approaches have attempted to solve this problem by implicitly modeling dependencies between outputs.In this paper, we introduce AligNART, which leverages full alignment information to explicitly reduce the modality of the target distribution.AligNART divides the machine translation task into (i) alignment estimation and (ii) translation with aligned decoder inputs, guiding the decoder to focus on simplified one-to-one translation.To alleviate the alignment estimation problem, we further propose a novel alignment decomposition method.Our experiments show that AligNART outperforms previous non-iterative NART models that focus on explicit modality reduction on WMT14 En↔De and WMT16 Ro→En.Furthermore, AligNART achieves BLEU scores comparable to those of the state-of-the-art connectionist temporal classification based models on WMT14 En↔De.We also observe that AligNART effectively addresses the token repetition problem even without sequence-level knowledge distillation.

Denoising diffusion probabilistic models (DDPM) have shown remarkable performance in unconditional image generation. However, due to the stochasticity of the generative process in DDPM, it is challenging to generate images with the desired semantics. In this work, we propose Iterative Latent Variable Refinement (ILVR), a method to guide the generative process in DDPM to generate high-quality images based on a given reference image. Here, the refinement of the generative process in DDPM enables a single DDPM to sample images from various sets directed by the reference image. The proposed ILVR method generates high-quality images while controlling the generation. The controllability of our method allows adaptation of a single DDPM without any additional learning in various image generation tasks, such as generation from various downsampling factors, multi-domain image translation, paint-to-image, and editing with scribbles.

Recent image generation models show remarkable generation performance. However, they mirror strong location preference in datasets, which we call spatial bias. Therefore, generators render poor samples at unseen locations and scales. We argue that the generators rely on their implicit positional encoding to render spatial content. From our observations, the generator’s implicit positional encoding is translation-variant, making the generator spatially biased. To address this issue, we propose injecting explicit positional encoding at each scale of the generator. By learning the spatially unbiased generator, we facilitate the robust use of generators in multiple tasks, such as GAN inversion, multi-scale generation, generation of arbitrary sizes and aspect ratios. Furthermore, we show that our method can also be applied to denoising diffusion probabilistic models. Our code is available at: https://github.com/jychoill8/toward_spatial_unbiased.

Deep neural networks continue to awe the world with their remarkable performance. Their predictions, however, are prone to be corrupted by adversarial examples that are imperceptible to humans. Current efforts to improve the robustness of neural networks against adversarial examples are focused on developing robust training methods, which update the weights of a neural network in a more robust direction. In this work, we take a step beyond training of the weight parameters and consider the problem of designing an adversarially robust neural architecture with high intrinsic robustness. We propose AdvRush, a novel adversarial robustness-aware neural architecture search algorithm, based upon a finding that independent of the training method, the intrinsic robustness of a neural network can be represented with the smoothness of its input loss landscape. Through a regularizer that favors a candidate architecture with a smoother input loss landscape, AdvRush successfully discovers an adversarially robust neural architecture. Along with a comprehensive theoretical motivation for AdvRush, we conduct an extensive amount of experiments to demonstrate the efficacy of AdvRush on various benchmark datasets. Notably, on CIFAR-10, AdvRush achieves 55.91% robust accuracy under FGSM attack after standard training and 50.04% robust accuracy under AutoAttack after 7-step PGD adversarial training.

Existing face authentication methods rely heavily on the general feature extractor trained with face verification algorithms without considering the target identity. No consideration of the target identity causes inefficiency in that the face authentication works on a specific target identity. Personalized face authentication systems that integrate target identity information is demanded to yield an efficient and superior performance than current methods. With the help of fast adaptation in few-shot meta-learning, we propose a novel personalized Face Authentication based on few-shot Meta-Learning (FAML). FAML introduces a unique person-specific binary classification task and trains an initialization model that can evolve into the personalized face authentication model for any target identity with few data and a small number of iterations. We evaluate the FAML on two public datasets, and it outperforms the baselines by a large margin and demonstrates the practicability as a security application.

Despite the increasing interest in neural architecture search (NAS), the significant computational cost of NAS is a hindrance to researchers. Hence, we propose to reduce the cost of NAS using proxy data, i.e., a representative subset of the target data, without sacrificing search performance. Even though data selection has been used across various fields, our evaluation of existing selection methods for NAS algorithms offered by NAS-Bench-1shot1 reveals that they are not always appropriate for NAS and a new selection method is necessary. By analyzing proxy data constructed using various selection methods through data entropy, we propose a novel proxy data selection method tailored for NAS. To empirically demonstrate the effectiveness, we conduct thorough experiments across diverse datasets, search spaces, and NAS algorithms. Consequently, NAS algorithms with the proposed selection discover architectures that are competitive with those obtained using the entire dataset. It significantly reduces the search cost: executing DARTS with the proposed selection requires only 40 minutes on CIFAR-10 and 7.5 hours on ImageNet with a single GPU. Additionally, when the architecture searched on ImageNet using the proposed selection is inversely transferred to CIFAR-10, a state-of-the-art test error of 2.4% is yielded. Our code is available at https://github.com/nabk89/NAS-with-Proxy-data.

Weakly supervised semantic segmentation produces pixel-level localization from class labels; however, a classifier trained on such labels is likely to focus on a small discriminative region of the target object. We interpret this phenomenon using the information bottleneck principle: the final layer of a deep neural network, activated by the sigmoid or softmax activation functions, causes an information bottleneck, and as a result, only a subset of the task-relevant information is passed on to the output. We first support this argument through a simulated toy experiment and then propose a method to reduce the information bottleneck by removing the last activation function. In addition, we introduce a new pooling method that further encourages the transmission of information from non-discriminative regions to the classification. Our experimental evaluations demonstrate that this simple modification significantly improves the quality of localization maps on both the PASCAL VOC 2012 and MS COCO 2014 datasets, exhibiting a new state-of-the-art performance for weakly supervised semantic segmentation.

MOTIVATION: Metagenomic sequencing has become a crucial tool for obtaining a gene catalogue of operational taxonomic units (OTUs) in a microbial community. A typical metagenomic sequencing produces a large amount of data (often in the order of terabytes or more), and computational tools are indispensable for efficient processing. In particular, error correction in metagenomics is crucial for accurate and robust genetic cataloging of microbial communities. However, many existing error-correction tools take a prohibitively long time and often bottleneck the whole analysis pipeline. RESULTS: To overcome this computational hurdle, we analyzed and exploited the data-level parallelism that exists in the error-correction procedure and proposed a tool named MUGAN that exploits both multi-core central processing units and multiple graphics processing units for co-processing. According to the experimental results, our approach reduced not only the time demand for denoising amplicons from approximately 59 h to only 46 min, but also the overestimation of the number of OTUs, estimating 6.7 times less species-level OTUs than the baseline. In addition, our approach provides web-based intuitive visualization of results. Given its efficiency and convenience, we anticipate that our approach would greatly facilitate denoising efforts in metagenomics studies. AVAILABILITY AND IMPLEMENTATION: http://data.snu.ac.kr/pub/mugan. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Memory-augmented neural networks (MANNs) were introduced to handle long-term dependent data efficiently. MANNs have shown promising results in question answering (QA) tasks that require holding contexts for answering a given question. As demands for QA on edge devices have increased, the utilization of MANNs in resource-constrained environments has become important. To achieve fast and energy-efficient inference of MANNs, we can exploit application-specific hardware accelerators on field-programmable gate arrays (FPGAs). Although several accelerators for conventional deep neural networks have been designed, it is difficult to efficiently utilize the accelerators with MANNs due to different requirements. In addition, characteristics of QA tasks should be considered for further improving the efficiency of inference on the accelerators. To address the aforementioned issues, we propose an inference accelerator of MANNs on FPGA. To fully utilize the proposed accelerator, we introduce fast inference methods considering the features of QA tasks. To evaluate our proposed approach, we implemented the proposed architecture on an FPGA and measured the execution time and energy consumption for the bAbI data set. According to our thorough experiments, the proposed methods improved speed and energy efficiency of the inference of MANNs up to about 25.6 and 28.4 times, respectively, compared with those of CPU.

Over the past decade, deep neural networks (DNNs) have demonstrated remarkable performance in a variety of applications. As we try to solve more advanced problems, increasing demands for computing and power resources has become inevitable. Spiking neural networks (SNNs) have attracted widespread interest as the third-generation of neural networks due to their event-driven and low-powered nature. SNNs, however, are difficult to train, mainly owing to their complex dynamics of neurons and non-differentiable spike operations. Furthermore, their applications have been limited to relatively simple tasks such as image classification. In this study, we investigate the performance degradation of SNNs in a more challenging regression problem (i.e., object detection). Through our in-depth analysis, we introduce two novel methods: channel-wise normalization and signed neuron with imbalanced threshold, both of which provide fast and accurate information transmission for deep SNNs. Consequently, we present a first spiked-based object detection model, called Spiking-YOLO. Our experiments show that Spiking-YOLO achieves remarkable results that are comparable (up to 98%) to those of Tiny YOLO on non-trivial datasets, PASCAL VOC and MS COCO. Furthermore, Spiking-YOLO on a neuromorphic chip consumes approximately 280 times less energy than Tiny YOLO and converges 2.3 to 4 times faster than previous SNN conversion methods.

Adversarial examples cause neural networks to produce incorrect outputs with high confidence. Although adversarial training is one of the most effective forms of defense against adversarial examples, unfortunately, a large gap exists between test accuracy and training accuracy in adversarial training. In this paper, we identify Adversarial Feature Overfitting (AFO), which may cause poor adversarially robust generalization, and we show that adversarial training can overshoot the optimal point in terms of robust generalization, leading to AFO in our simple Gaussian model. Considering these theoretical results, we present soft labeling as a solution to the AFO problem. Furthermore, we propose Adversarial Vertex mixup (AVmixup), a soft-labeled data augmentation approach for improving adversarially robust generalization. We complement our theoretical analysis with experiments on CIFAR10, CIFAR100, SVHN, and Tiny ImageNet, and show that AVmixup significantly improves the robust generalization performance and that it reduces the trade-off between standard accuracy and adversarial robustness.

Spiking neural networks (SNNs) have gained considerable interest due to their energy-efficient characteristics, yet lack of a scalable training algorithm has restricted their applicability in practical machine learning problems. The deep neural network-to-SNN conversion approach has been widely studied to broaden the applicability of SNNs. Most previous studies, however, have not fully utilized spatio-temporal aspects of SNNs, which has led to inefficiency in terms of number of spikes and inference latency. In this paper, we present T2FSNN, which introduces the concept of time-to-first-spike coding into deep SNNs using the kernel-based dynamic threshold and dendrite to overcome the aforementioned drawback. In addition, we propose gradient-based optimization and early firing methods to further increase the efficiency of the T2FSNN. According to our results, the proposed methods can reduce inference latency and number of spikes to 22% and less than 1%, compared to those of burst coding, which is the state-of-the-art result on the CIFAR-100.

Field-programmable gate arrays (FPGAs) have been widely used in many applications due to their reconfigurability. Notably, the short development time makes the FPGAs one of the promising reconfigurable architectures for emerging applications, such as deep learning. As CMOS technology advances, however, conventional SRAM-based FPGAs have reached their limitations. To overcome these obstacles, NVM-based FPGAs have been introduced. Although NVM-based FPGAs have the features of high area density, low static power consumption, and non-volatility, they are struggling to reduce energy consumption. Their challenge is mainly caused by the access speed of NVM, which is relatively slower than SRAM. In this paper, for compensating this limitation, we suggest SRAM-NVM hybrid FPGA architecture with SRAM- and NVM-based CLBs. In addition, we propose an energy-aware placement for utilizing the SRAM-NVM hybrid FPGAs. As a result of our experiments, we were able to reduce the average energy consumption of SRAM-NVM hybrid FPGA by 22.23% and 21.94% compared to SRAM-based FPGA on the MCNC and VTR benchmarks, respectively.

To demystify the "black box" property of deep neural networks for natural language processing (NLP), several methods have been proposed to interpret their predictions by measuring the change in prediction probability after erasing each token of an input.Since existing methods replace each token with a predefined value (i.e., zero), the resulting sentence lies out of the training data distribution, yielding misleading interpretations.In this study, we raise the out-of-distribution problem induced by the existing interpretation methods and present a remedy; we propose to marginalize each token out.We interpret various NLP models trained for sentiment analysis and natural language inference using the proposed method.

Recently, text-to-speech (TTS) models such as FastSpeech and ParaNet have been proposed to generate mel-spectrograms from text in parallel. Despite the advantage, the parallel TTS models cannot be trained without guidance from autoregressive TTS models as their external aligners. In this work, we propose Glow-TTS, a flow-based generative model for parallel TTS that does not require any external aligner. By combining the properties of flows and dynamic programming, the proposed model searches for the most probable monotonic alignment between text and the latent representation of speech on its own. We demonstrate that enforcing hard monotonic alignments enables robust TTS, which generalizes to long utterances, and employing generative flows enables fast, diverse, and controllable speech synthesis. Glow-TTS obtains an order-of-magnitude speed-up over the autoregressive model, Tacotron 2, at synthesis with comparable speech quality. We further show that our model can be easily extended to a multi-speaker setting.

Normalizing flows (NFs) have become a prominent method for deep generative models that allow for an analytic probability density estimation and efficient synthesis. However, a flow-based network is considered to be inefficient in parameter complexity because of reduced expressiveness of bijective mapping, which renders the models unfeasibly expensive in terms of parameters. We present an alternative parameterization scheme called NanoFlow, which uses a single neural density estimator to model multiple transformation stages. Hence, we propose an efficient parameter decomposition method and the concept of flow indication embedding, which are key missing components that enable density estimation from a single neural network. Experiments performed on audio and image models confirm that our method provides a new parameter-efficient solution for scalable NFs with significant sublinear parameter complexity.

Drugs are classified according to their biological and chemical reactions, and the systems that they target. Thus, an accurate and efficient prediction method for drug class discovery would reveal key properties of candidate drugs, significantly conserving time and resources in drug repositioning and design. Previous approaches, based on data mining or statistics, required complicated feature construction in advance. Knowing that deep learning can identifying patterns in high-dimensional datasets without elaborate feature selection or engineering, we constructed a model for predicting drug classes using deep neural networks - with biological and chemical structure data. Our proposed model outperforms previous learning-based methods in terms of prediction accuracy.

The main obstacle to weakly supervised semantic image segmentation is the difficulty of obtaining pixel-level information from coarse image-level annotations. Most methods based on image-level annotations use localization maps obtained from the classifier, but these only focus on the small discriminative parts of objects and do not capture precise boundaries. FickleNet explores diverse combinations of locations on feature maps created by generic deep neural networks. It selects hidden units randomly and then uses them to obtain activation scores for image classification. FickleNet implicitly learns the coherence of each location in the feature maps, resulting in a localization map which identifies both discriminative and other parts of objects. The ensemble effects are obtained from a single network by selecting random hidden unit pairs, which means that a variety of localization maps are generated from a single image. Our approach does not require any additional training steps and only adds a simple layer to a standard convolutional neural network; nevertheless it outperforms recent comparable techniques on the Pascal VOC 2012 benchmark in both weakly and semi-supervised settings.

Spiking neural networks (SNNs) are considered as one of the most promising artificial neural networks due to their energy-efficient computing capability. Recently, conversion of a trained deep neural network to an SNN has improved the accuracy of deep SNNs. However, most of the previous studies have not achieved satisfactory results in terms of inference speed and energy efficiency. In this paper, we propose a fast and energy-efficient information transmission method with burst spikes and hybrid neural coding scheme in deep SNNs. Our experimental results showed the proposed methods can improve inference energy efficiency and shorten the latency.

Attacks on embedded devices are likely to occur any time in unexpected manners. Thus, the defense systems based on fixed sets of rules will easily be subverted by such unexpected, unknown attacks. Learning-based anomaly detection may potentially prevent new unknown zero-day attacks by leveraging the capability of machine learning (ML) to learn the intricate true nature of software hidden within raw information. This paper introduces our work to develop an MPSoC, called RTAD, which can efficiently support in hardware various ML models that run to detect anomalous behaviors on embedded devices in a real-time fashion, and thus enable the devices to counteract the anomalies in the field. In the IoT era, the importance of security for embedded devices cannot be exaggerated because they will become an enticing target for adversaries as they are being integrated into everyday life to provide users with various services. The above-mentioned potential of learning-based detection is believed to benefit those deployed devices under attacks occurring any time during their field operations in unexpected manners. We hereby assume that ML models are trained with runtime branch information as their data features since a sequence of branches serves as a record of control flow transfers during program execution. In fact, there have been numerous ML studies that examine various types of branches in order to infer (or detect) anomaly in branch behaviors that may be induced by diverse attacks that can cause deviant control flow in software. Our goal of real-time anomalous branch behavior inference poses two challenges to our development of RTAD. Firstly, RTAD must collect and transfer in a timely fashion a sequence of branches as the input to the ML model. Secondly, RTAD must be able to promptly process the delivered branch data with the ML model. To tackle these challenges, we have implemented in RTAD two core components: an input generation module and a GPU-inspired ML processing engine. According to our experiments, RTAD enables various ML models to infer anomaly instantly after the victim program behaves aberrantly as the result of attacks being injected into the system.

Memory-augmented neural networks (MANNs) are designed for question-answering tasks. It is difficult to run a MANN effectively on accelerators designed for other neural networks (NNs), in particular on mobile devices, because MANNs require recurrent data paths and various types of operations related to external memory access. We implement an accelerator for MANNs on a field-programmable gate array (FPGA) based on a data flow architecture. Inference times are also reduced by inference thresholding, which is a data-based maximum inner-product search specialized for natural language tasks. Measurements on the bAbI data show that the energy efficiency of the accelerator (FLOPS/kJ) was higher than that of an NVIDIA TITAN V GPU by a factor of about 125, increasing to 140 with inference thresholding.

When a deep neural network is trained on data with only image-level labeling, the regions activated in each image tend to identify only a small region of the target object. We propose a method of using videos automatically harvested from the web to identify a larger region of the target object by using temporal information, which is not present in the static image. The temporal variations in a video allow different regions of the target object to be activated. We obtain an activated region in each frame of a video, and then aggregate the regions from successive frames into a single image, using a warping technique based on optical flow. The resulting localization maps cover more of the target object, and can then be used as proxy ground-truth to train a segmentation network. This simple approach outperforms existing methods under the same level of supervision, and even approaches relying on extra annotations. Based on VGG-16 and ResNet 101 backbones, our method achieves the mIoU of 65.0 and 67.4, respectively, on PASCAL VOC 2012 test images, which represents a new state-of-the-art.

Most deep learning classification studies assume clean data. However, when dealing with the real world data, we encounter three problems such as 1) missing data, 2) class imbalance, and 3) missing label problems. These problems undermine the performance of a classifier. Various preprocessing techniques have been proposed to mitigate one of these problems, but an algorithm that assumes and resolves all three problems together has not been proposed yet. In this paper, we propose HexaGAN, a generative adversarial network framework that shows promising classification performance for all three problems. We interpret the three problems from a single perspective to solve them jointly. To enable this, the framework consists of six components, which interact with each other. We also devise novel loss functions corresponding to the architecture. The designed loss functions allow us to achieve state-of-the-art imputation performance, with up to a 14% improvement, and to generate high-quality class-conditional data. We evaluate the classification performance (F1-score) of the proposed method with 20% missingness and confirm up to a 5% improvement in comparison with the performance of combinations of state-of-the-art methods.

Most modern text-to-speech architectures use a WaveNet vocoder for synthesizing high-fidelity waveform audio, but there have been limitations, such as high inference time, in practical applications due to its ancestral sampling scheme. The recently suggested Parallel WaveNet and ClariNet has achieved real-time audio synthesis capability by incorporating inverse autoregressive flow (IAF) for parallel sampling. However, these approaches require a two-stage training pipeline with a well-trained teacher network and can only produce natural sound by using probability distillation along with heavily-engineered auxiliary loss terms. We propose FloWaveNet, a flow-based generative model for raw audio synthesis. FloWaveNet requires only a single-stage training procedure and a single maximum likelihood loss, without any additional auxiliary terms, and it is inherently parallel due to the characteristics of generative flow. The model can efficiently sample raw audio in real-time, with clarity comparable to previous two-stage parallel models. The code and samples for all models, including our FloWaveNet, are available on GitHub.

BACKGROUND: Quantitative structure-activity relationship (QSAR) is a computational modeling method for revealing relationships between structural properties of chemical compounds and biological activities. QSAR modeling is essential for drug discovery, but it has many constraints. Ensemble-based machine learning approaches have been used to overcome constraints and obtain reliable predictions. Ensemble learning builds a set of diversified models and combines them. However, the most prevalent approach random forest and other ensemble approaches in QSAR prediction limit their model diversity to a single subject. RESULTS: The proposed ensemble method consistently outperformed thirteen individual models on 19 bioassay datasets and demonstrated superiority over other ensemble approaches that are limited to a single subject. The comprehensive ensemble method is publicly available at http://data.snu.ac.kr/QSAR/ . CONCLUSIONS: We propose a comprehensive ensemble method that builds multi-subject diversified models and combines them through second-level meta-learning. In addition, we propose an end-to-end neural network-based individual classifier that can automatically extract sequential features from a simplified molecular-input line-entry system (SMILES). The proposed individual models did not show impressive results as a single model, but it was considered the most important predictor when combined, according to the interpretation of the meta-learning.

Chemical-chemical interaction (CCI) plays a major role in predicting candidate drugs, toxicities, therapeutic effects, and biological functions. CCI is typically inferred from a variety of information; however, CCI has yet not been predicted using a learning-based approach. In other drug analyses, deep learning has been actively used in recent years. However, in most cases, deep learning has been used only for classification even though it has feature extraction capabilities. Thus, in this paper, we propose an end-to-end representation learning method for CCI, named DeepCCI, which includes feature extraction and a learning-based approach. Our proposed architecture is based on the Siamese network. Hidden representations are extracted from a simplified molecular input line entry system (SMILES), which is a string notation representing the chemical structure using weight-shared convolutional neural networks. Subsequently, L1 element-wise distances between the two extracted hidden representations are measured. The performance of DeepCCI is compared with those of 12 fingerprint-method combinations. The proposed DeepCCI shows the best performance in most of the evaluation metrics used. In addition, DeepCCI was experimentally validated to guarantee the commutative property. The automatically extracted features can alleviate the efforts required for manual feature engineering and improve prediction performance.

Memory-augmented neural networks (MANNs) refer to a class of neural network models equipped with external memory (such as neural Turing machines and memory networks). These neural networks outperform conventional recurrent neural networks (RNNs) in terms of learning long-term dependency, allowing them to solve intriguing AI tasks that would otherwise be hard to address. This paper concerns the problem of quantizing MANNs. Quantization is known to be effective when we deploy deep models on embedded systems with limited resources. Furthermore, quantization can substantially reduce the energy consumption of the inference procedure. These benefits justify recent developments of quantized multi layer perceptrons, convolutional networks, and RNNs. However, no prior work has reported the successful quantization of MANNs. The in-depth analysis presented here reveals various challenges that do not appear in the quantization of the other networks. Without addressing them properly, quantized MANNs would normally suffer from excessive quantization error which leads to degraded performance. In this paper, we identify memory addressing (specifically, content-based addressing) as the main reason for the performance degradation and propose a robust quantization method for MANNs to address the challenge. In our experiments, we achieved a computation-energy gain of 22× with 8-bit fixed-point and binary quantization compared to the floating-point implementation. Measured on the bAbI dataset, the resulting model, named the quantized MANN (Q-MANN), improved the error rate by 46% and 30% with 8-bit fixed-point and binary quantization, respectively, compared to the MANN quantized using conventional techniques.

Detecting major system anomalies with observed multivariate time series requires not only the characteristics of each time series but also the status of the entire time series dynamics. Therefore, we propose a method that can detect substantial anomalies by generating a transfer network and an influence network from a multivariate time series. To form a transfer network, each vertex represents a single time series. Each edge indicates the strength of the information flow between each pair of time series using transfer entropy. With the transfer network, we exploit the random walk approach to calculate the affinity score between two vertices and create an influence network that reflects both the direct and indirect influences. In our experiment, we show the efficacy of the proposed method using simple synthetic time series networks and the real data set such as world stock indices and key performance indicators of the SAP HANA in-memory database system.

Motivation: Long non-coding RNAs (lncRNAs) are important regulatory elements in biological processes. LncRNAs share similar sequence characteristics with messenger RNAs, but they play completely different roles, thus providing novel insights for biological studies. The development of next-generation sequencing has helped in the discovery of lncRNA transcripts. However, the experimental verification of numerous transcriptomes is time consuming and costly. To alleviate these issues, a computational approach is needed to distinguish lncRNAs from the transcriptomes. Results: We present a deep learning-based approach, lncRNAnet, to identify lncRNAs that incorporates recurrent neural networks for RNA sequence modeling and convolutional neural networks for detecting stop codons to obtain an open reading frame indicator. lncRNAnet performed clearly better than the other tools for sequences of short lengths, on which most lncRNAs are distributed. In addition, lncRNAnet successfully learned features and showed 7.83%, 5.76%, 5.30% and 3.78% improvements over the alternatives on a human test set in terms of specificity, accuracy, F1-score and area under the curve, respectively. Availability and implementation: Data and codes are available in http://data.snu.ac.kr/pub/lncRNAnet.

BACKGROUND: DNA damage causes aging, cancer, and other serious diseases. The comet assay can detect multiple types of DNA lesions with high sensitivity, and it has been widely applied. Although comet assay platforms have improved the limited throughput and reproducibility of traditional assays in recent times, analyzing large quantities of comet data often requires a tremendous human effort. To overcome this challenge, we proposed HiComet, a computational tool that can rapidly recognize and characterize a large number of comets, using little user intervention. RESULTS: We tested HiComet with real data from 35 high-throughput comet assay experiments, with over 700 comets in total. The proposed method provided unprecedented levels of performance as an automated comet recognition tool in terms of robustness (measured by precision and recall) and throughput. CONCLUSIONS: HiComet is an automated tool for high-throughput comet-assay analysis and could significantly facilitate characterization of individual comets by accelerating its most rate-limiting step. An online implementation of HiComet is freely available at https://github.com/taehoonlee/HiComet/ .

After publication of the original article [1], it has been found that the author affiliations have been accidentally left out in the PDF. The full affiliations can be found in this correction.

To assess the genetic diversity of an environmental sample in metagenomics studies, the amplicon sequences of 16s rRNA genes need to be clustered into operational taxonomic units (OTUs). Many existing tools for OTU clustering trade off between accuracy and computational efficiency. We propose a novel OTU clustering algorithm, hc-OTU, which achieves high accuracy and fast runtime by exploiting homopolymer compaction and k-mer profiling to significantly reduce the computing time for pairwise distances of amplicon sequences. We compare the proposed method with other widely used methods, including UCLUST, CD-HIT, MOTHUR, ESPRIT, ESPRIT-TREE, and CLUSTOM, comprehensively, using nine different experimental datasets and many evaluation metrics, such as normalized mutual information, adjusted Rand index, measure of concordance, and F-score. Our evaluation reveals that the proposed method achieves a level of accuracy comparable to the respective accuracy levels of MOTHUR and ESPRIT-TREE, two widely used OTU clustering methods, while delivering orders-of-magnitude speedups.

Graphs provide a powerful means for representing complex interactions between entities. Recently, new deep learning approaches have emerged for representing and modeling graph-structured data while the conventional deep learning methods, such as convolutional neural networks and recurrent neural networks, have mainly focused on the grid-structured inputs of image and audio. Leveraged by representation learning capabilities, deep learning-based techniques can detect structural characteristics of graphs, giving promising results for graph applications. In this paper, we attempt to advance deep learning for graph-structured data by incorporating another component: transfer learning. By transferring the intrinsic geometric information learned in the source domain, our approach can construct a model for a new but related task in the target domain without collecting new data and without training a new model from scratch. We thoroughly tested our approach with large-scale real-world text data and confirmed the effectiveness of the proposed transfer learning framework for deep learning on graphs. According to our experiments, transfer learning is most effective when the source and target domains bear a high level of structural similarity in their graph representations.

MicroRNAs (miRNAs) are small non-coding ribonucleic acids (RNAs) which play key roles in post-transcriptional gene regulation. Direct identification of mature miRNAs is infeasible due to their short lengths, and researchers instead aim at identifying precursor miRNAs (pre-miRNAs). Many of the known pre-miRNAs have distinctive stem-loop secondary structure, and structure-based filtering is usually the first step to predict the possibility of a given sequence being a pre-miRNA. To identify new pre-miRNAs that often have non-canonical structure, however, we need to consider additional features other than structure. To obtain such additional characteristics, existing computational methods rely on manual feature extraction, which inevitably limits the efficiency, robustness, and generalization of computational identification. To address the limitations of existing approaches, we propose a pre-miRNA identification method that incorporates (1) a deep recurrent neural network (RNN) for automated feature learning and classification, (2) multimodal architecture for seamless integration of prior knowledge (secondary structure), (3) an attention mechanism for improving long-term dependence modeling, and (4) an RNN-based class activation mapping for highlighting the learned representations that can contrast pre-miRNAs and non-pre-miRNAs. In our experiments with recent benchmarks, the proposed approach outperformed the compared state-of-the-art alternatives in terms of various performance metrics.

In the era of big data, transformation of biomedical big data into valuable knowledge has been one of the most important challenges in bioinformatics. Deep learning has advanced rapidly since the early 2000s and now demonstrates state-of-the-art performance in various fields. Accordingly, application of deep learning in bioinformatics to gain insight from data has been emphasized in both academia and industry. Here, we review deep learning in bioinformatics, presenting examples of current research. To provide a useful and comprehensive perspective, we categorize research both by the bioinformatics domain (i.e. omics, biomedical imaging, biomedical signal processing) and deep learning architecture (i.e. deep neural networks, convolutional neural networks, recurrent neural networks, emergent architectures) and present brief descriptions of each study. Additionally, we discuss theoretical and practical issues of deep learning in bioinformatics and suggest future research directions. We believe that this review will provide valuable insights and serve as a starting point for researchers to apply deep learning approaches in their bioinformatics studies.

Today's datacenters are equipped with diverse computing and storage devices for handling a myriad of data and normally consume a significant amount of electrical energy. This paper proposes a smart grid inspired methodology to monitor and profile the energy consumption of a datacenter, with the aim of providing information useful for reducing the peak power consumption of the datacenter. Our energy measurement platform is named CloudSocket, and each CloudSocket unit can measure the power consumption of an individual computing node and periodically transmit the measurement information wirelessly to the coordinator unit that can manage many Cloud-Sockets simultaneously. We tested our methodology with a 32-node grid system that runs Apache Spark for large-scale data analytics. Analyzing our experimental results reveals how and where the peak power of each node in the grid overlaps, providing opportunities for informative coordination of the computing components for overall power reduction.

As a great deal of data has been accumulated in various disciplines, the need for the integrative analysis of separate but relevant data sources is becoming more important. Combining data sources can provide global insight that is otherwise difficult to obtain from individual sources. Because of privacy, regulations, and other issues, many large-scale data repositories remain closed off from the outside, raising what has been termed the data silo issue. The huge volume of today's big data often leads to computational challenges, adding another layer of complexity to the solution. In this paper, we propose a novel method called collaborative analytics by ensemble learning (CABEL), which attempts to resolve the main hurdles regarding the silo issue: accuracy, privacy, and computational efficiency. CABEL represents the data stored in each silo as a compact aggregate of samples called the silo signature. The compact representation provides computational efficiency and privacy preservation but makes it challenging to produce accurate analytics. To resolve this challenge, we formulate the problem of attribute domain sampling and reconstruction, and propose a solution called the Chebyshev subset. To model collaborative efforts to analyze semantically linked but structurally disconnected databases, CABEL utilizes a new ensemble learning technique termed the weighted bagging of base classifiers. We demonstrate the effectiveness of CABEL by testing with a nationwide health-insurance data set containing approximately 4,182,000,000 records collected from the entire population of an Organisation for Economic Co-operation and Development (OECD) country in 2012. In our binary classification tests, CABEL achieved median recall, precision, and F-measure values of 89%, 64%, and 76%, respectively, although only 0.001–0.00001% of the original data was used for model construction, while maintaining data privacy and computational efficiency.

We present a new framework of applying deep neural networks (DNN) to devise a universal discrete denoiser. Unlike other approaches that utilize supervised learning for denoising, we do not require any additional training data. In such setting, while the ground-truth label, i.e., the clean data, is not available, we devise ``pseudo-labels'' and a novel objective function such that DNN can be trained in a same way as supervised learning to become a discrete denoiser. We experimentally show that our resulting algorithm, dubbed as Neural DUDE, significantly outperforms the previous state-of-the-art in several applications with a systematic rule of choosing the hyperparameter, which is an attractive feature in practice.

A wide variety of large-scale data have been produced in bioinformatics. In response, the need for efficient handling of biomedical big data has been partly met by parallel computing. However, the time demand of many bioinformatics programs still remains high for large-scale practical uses because of factors that hinder acceleration by parallelization. Recently, new generations of storage devices have emerged, such as NAND flash-based solid-state drives (SSDs), and with the renewed interest in near-data processing, they are increasingly becoming acceleration methods that can accompany parallel processing. In certain cases, a simple drop-in replacement of hard disk drives by SSDs results in dramatic speedup. Despite the various advantages and continuous cost reduction of SSDs, there has been little review of SSD-based profiling and performance exploration of important but time-consuming bioinformatics programs. For an informative review, we perform in-depth profiling and analysis of 23 key bioinformatics programs using multiple types of devices. Based on the insight we obtain from this research, we further discuss issues related to design and optimize bioinformatics algorithms and pipelines to fully exploit SSDs. The programs we profile cover traditional and emerging areas of importance, such as alignment, assembly, mapping, expression analysis, variant calling and metagenomics. We explain how acceleration by parallelization can be combined with SSDs for improved performance and also how using SSDs can expedite important bioinformatics pipelines, such as variant calling by the Genome Analysis Toolkit and transcriptome analysis using RNA sequencing. We hope that this review can provide useful directions and tips to accompany future bioinformatics algorithm design procedures that properly consider new generations of powerful storage devices.

Recovery from sudden power-off (SPO) is one of the primary concerns among practitioners which bars the quick and wide deployment of flash storage devices. In this work, we propose Metadata Embedded Write (MEW), a novel scheme for handling the sudden power-off recovery in modern flash storage devices. Given that a large fraction of commercial SSDs employ compression technology, MEW exploits the compression-induced internal fragmentation in the data area to store rich metadata for fast and complete recovery. MEW consists of (i) a metadata embedding scheme to harbor SSD metadata in a physical page together with multiple compressed logical pages, (ii) an allocation chain based fast recovery scheme, and (iii) a light-weight metadata logging scheme which enables MEW to maintain the metadata for incompressible data, too. We performed extensive experiments to examine the performance of MEW. The performance overhead of MEW is 3 percent in the worst case, in terms of the write amplification factor, compared to the pure compression-based FTL that does not have any recovery scheme.

In recent years, the demand for NAND flash-based storage devices has rapidly increased because of the popularization of various portable devices. NAND flash memory (NFM) offers many advantages, such as nonvolatility, high performance, the small form factor, and low-power consumption, while achieving high chip integration with a specialized architecture for bulk data access. A unit of NFM's read and program operations, the page, has continuously grown. Although increasing page size reduces costs, it adversely affects performance because of the resultant side effects, such as fragmentation and wasted space, caused by the incongruity of data and page sizes. To address this issue, we propose a multiple-page-size NFM architecture and its management. Our method dramatically improves write performance through adopting multiple page sizes without requiring additional area overhead or manufacturing processes. Based on the experimental results, the proposed NFM improves write latency and NFM lifetime by up to 65% and 62%, respectively, compared with the single-page-size NFM.

By how much is a large-scale graph transformed over time or by a significant event?' or 'how structurally similar are two large-scale graphs?' are the two questions that this paper attempts to address. The proposed method efficiently calculates and accurately produces graph similarity. Our approach is based on the well-known random walk with restart (RWR) algorithm, which quantifies relevance between nodes to express the structural and connection characteristics of graphs. Intergraph compression, which is inspired by interframe compression, merges two input graphs and reorders their nodes contributing to improved process-data storage efficiency and processing convenience. This is a boon to the RWR algorithm for large-scale graphs. The representation of a graph transformed via intergraph compression can be used to accurately show similarity because sub-matrix blocks are reordered to concentrate nonzero elements. In performing the RWR algorithm, which quantifies inter-node relevance, transformed representation of graph with Intergraph compression is efficient in space requirement and produces results more quickly and accurately over conventional graph transformation schemes. We demonstrate the validity of our method through experiments and apply it to the usage data of public transportation SmartCard in a large metropolitan area to suggest usefulness of the proposed algorithm.

Splicing refers to the elimination of non-coding regions in transcribed pre-messenger ribonucleic acid (RNA). Discovering splice sites is an important machine learning task that helps us not only to identify the basic units of genetic heredity but also to understand how different proteins are produced. Existing methods for splicing prediction have produced promising results, but often show limited robustness and accuracy. In this paper, we propose a deep belief network-based methodology for computational splice junction prediction. Our proposal includes a novel method for training restricted Boltzmann machines for class-imbalanced prediction. The proposed method addresses the limitations of conventional contrastive divergence and provides regularization for datasets that have categorical features. We tested our approach using public human genome datasets and obtained significantly improved accuracy and reduced runtime compared to state-of-the-art alternatives. The proposed approach was less sensitive to the length of input sequences and more robust for handling false splicing signals. Furthermore, we could discover non-canonical splicing patterns that were otherwise difficult to recognize using conventional methods. Given the efficiency and robustness of our methodology, we anticipate that it can be extended to the discovery of primary structural patterns of other subtle genomic elements.

MOTIVATION: Capillary electrophoresis (CE) is a powerful approach for structural analysis of nucleic acids, with recent high-throughput variants enabling three-dimensional RNA modeling and the discovery of new rules for RNA structure design. Among the steps composing CE analysis, the process of finding each band in an electrophoretic trace and mapping it to a position in the nucleic acid sequence has required significant manual inspection and remains the most time-consuming and error-prone step. The few available tools seeking to automate this band annotation have achieved limited accuracy and have not taken advantage of information across dozens of profiles routinely acquired in high-throughput measurements. RESULTS: We present a dynamic-programming-based approach to automate band annotation for high-throughput capillary electrophoresis. The approach is uniquely able to define and optimize a robust target function that takes into account multiple CE profiles (sequencing ladders, different chemical probes, different mutants) collected for the RNA. Over a large benchmark of multi-profile datasets for biological RNAs and designed RNAs from the EteRNA project, the method outperforms prior tools (QuSHAPE and FAST) significantly in terms of accuracy compared with gold-standard manual annotations. The amount of computation required is reasonable at a few seconds per dataset. We also introduce an 'E-score' metric to automatically assess the reliability of the band annotation and show it to be practically useful in flagging uncertainties in band annotation for further inspection. AVAILABILITY AND IMPLEMENTATION: The implementation of the proposed algorithm is included in the HiTRACE software, freely available as an online server and for download at http://hitrace.stanford.edu. CONTACT: [email protected] or [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In this work, we studied the energy consumption characteristics of various SSD design parameters. We developed an accurate energy consumption model for SSDs that computes aggregate, as well as component-specific, energy consumption of SSDs in sub-msec time scale. In our study, we used five different FTLs (page mapping, DFTL, block mapping, and two different hybrid mappings) and four different channel configurations (two, four, eight, and 16 channels) under seven different workloads (from large-scale enterprise systems to small-scale desktop applications) in a combinatorial manner. For each combination of the aforementioned parameters, we examined the energy consumption for individual hardware components of an SSD (microcontroller, DRAM, NAND flash, and host interface). The following are some of our findings. First, DFTL is the most energy-efficient address-mapping scheme among the five FTLs we tested due to its good write amplification and small DRAM footprint. Second, a significant fraction of energy is being consumed by idle flash chips waiting for the completion of NAND operations in the other channels. FTL should be designed to fully exploit the internal parallelism so that energy consumption by idle chips is minimized. Third, as a means to increase the internal parallelism, increasing way parallelism (the number of flash chips in a channel) is more effective than increasing channel parallelism in terms of peak energy consumption, performance, and hardware complexity. Fourth, in designing high-performance and energy-efficient SSDs, channel switching delay, way switching delay, and page write latency need to be incorporated in an integrated manner to determine the optimal configuration of internal parallelism.

Hierarchical agglomerative clustering (HAC) is a clustering method widely used in various disciplines from astronomy to zoology. HAC is useful for discovering hierarchical structure embedded in input data. The cost of executing HAC on large data is typically high, due to the need for maintaining global inter-cluster distance information throughout the execution. To address this issue, we propose a new parallelization scheme for multi-threaded shared-memory machines based on the concept of nearest-neighbor (NN) chains. The proposed multi-threaded algorithm allocates available threads into two groups, one for managing NN chains and the other for updating distance information. In-depth analysis of our approach gives insight into the ideal configuration of threads and theoretical performance bounds. We evaluate our proposed method by testing it with multiple public datasets and comparing its performance with that of several alternatives. In our test, the proposed method completes hierarchical clustering 3.09-51.79 times faster than the alternatives. Our test results also reveal the effects of performance-limiting factors such as starvation in chain growing, overhead incurred from using synchronization locks, and hardware aspects including memory-bandwidth saturation. According to our evaluation, the proposed scheme is effective in improving the HAC algorithm, achieving significant gains over the alternatives in terms of runtime and scalability.

Merging the forward and reverse reads from paired-end sequencing is a critical task that can significantly improve the performance of downstream tasks, such as genome assembly and mapping, by providing them with virtually elongated reads. However, due to the inherent limitations of most paired-end sequencers, the chance of observing erroneous bases grows rapidly as the end of a read is approached, which becomes a critical hurdle for accurately merging paired-end reads. Although there exist several sophisticated approaches to this problem, their performance in terms of quality of merging often remains unsatisfactory. To address this issue, here we present a context-aware scheme for paired-end reads (CASPER): a computational method to rapidly and robustly merge overlapping paired-end reads. Being particularly well suited to amplicon sequencing applications, CASPER is thoroughly tested with both simulated and real high-throughput amplicon sequencing data. According to our experimental results, CASPER significantly outperforms existing state-of-the art paired-end merging tools in terms of accuracy and robustness. CASPER also exploits the parallelism in the task of paired-end merging and effectively speeds up by multithreading. CASPER is freely available for academic use at http://best.snu.ac.kr/casper.

Over the past few decades, IO workload characterization has been a critical issue for operating system and storage community. Even so, the issue still deserves investigation because of the continued introduction of novel storage devices such as solid-state drives (SSDs), which have different characteristics from traditional hard disks. We propose novel IO workload characterization and classification schemes, aiming at addressing three major issues: (i) deciding right mining algorithms for IO traffic analysis, (ii) determining a feature set to properly characterize IO workloads, and (iii) defining essential IO traffic classes state-of-the-art storage devices can exploit in their internal management. The proposed characterization scheme extracts basic attributes that can effectively represent the characteristics of IO workloads and, based on the attributes, finds representative access patterns in general workloads using various clustering algorithms. The proposed classification scheme finds a small number of representative patterns of a given workload that can be exploited for optimization either in the storage stack of the operating system or inside the storage device.

In this paper, we present a virtual machine based SSD Simulator, VSSIM (Virtual SSD Simulator). VSSIM intends to address the issues of the trace driven simulation, e.g. trace re-scaling, accurate replay, etc. VSSIM operates on top of QEMU/KVM with software based SSD module. VSSIM runs in realtime and allows the user to measure both the host performance and the SSD behavior under various design choices. VSSIM can flexibly model the various hardware components, e.g. the number of channels, the number of ways, block size, page size, planes per chip, program, erase, read latency of NAND cells, channel switch delay, and way switch delay. VSSIM can also facilitate the implementation of the SSD firmware algorithms. To demonstrate the capability of VSSIM, we performed a number of case studies. The results of the simulation study deliver an important guideline in the firmware and hardware designs of future NAND based storage devices. Followings are some of the findings: (i) as the page size increases, the performance benefit of increasing the channel parallelism against increasing the way parallelism becomes less significant, (ii) due to the bi-modality in IO size distribution, FTL should be designed to handle multiple mapping granularity, (iii) hybrid mapping does not work in four or more way SSD due to severe log block fragmentation, (iv) as a performance metric, the Write Amplification Factor can be misleading, (v) compared to sequential write, random write operation can be benefited more from the channel level parallelism and therefore in multi-channel environment, it is beneficial to categorize larger fraction of IO as random. VSSIM is validated against commodity SSD, Intel X25M SSD. VSSIM models the sequential IO performance of X25M within 3% offset.

Direct projection of 3D branching structures, such as networks of cables, blood vessels, or neurons onto a 2D image creates the illusion of intersecting structural parts and creates challenges for understanding and communication. We present a method for visualizing such structures, and demonstrate its utility in visualizing the abdominal aorta and its branches, whose tomographic images might be obtained by computed tomography or magnetic resonance angiography, in a single 2D stylistic image, without overlaps among branches. The visualization method, termed uncluttered single-image visualization (USIV), involves optimization of geometry. This paper proposes a novel optimization technique that utilizes an interesting connection of the optimization problem regarding USIV to the protein structure prediction problem. Adopting the integer linear programming-based formulation for the protein structure prediction problem, we tested the proposed technique using 30 visualizations produced from five patient scans with representative anatomical variants in the abdominal aortic vessel tree. The novel technique can exploit commodity-level parallelism, enabling use of general-purpose graphics processing unit (GPGPU) technology that yields a significant speedup. Comparison of the results with the other optimization technique previously reported elsewhere suggests that, in most aspects, the quality of the visualization is comparable to that of the previous one, with a significant gain in the computation time of the algorithm.

Metagenomic sequencing has become a crucial tool for obtaining a gene catalogue of operational taxonomic units (OTUs) in a microbial community. High-throughput pyrosequencing is a next-generation sequencing technique very popular in microbial community analysis due to its longer read length compared to alternative methods. Computational tools are inevitable to process raw data from pyrosequencers, and in particular, noise removal is a critical data-mining step to obtain robust sequence reads. However, the slow rate of existing denoisers has bottlenecked the whole pyrosequencing process, let alone hindering efforts to improve robustness. To address these, we propose a new approach that can accelerate the denoising process substantially. By using our approach, it now takes only about 2 hours to denoise 62,873 pyrosequenced amplicons from a mixture of 91 full-length 16S rRNA clones. It would otherwise take nearly 2.5 days if existing software tools were used. Furthermore, our approach can effectively reduce overestimating the number of OTUs, producing 6.7 times fewer species-level OTUs on average than a state-of-the-art alternative under the same condition. Leveraged by our approach, we hope that metagenomic sequencing will become an even more appealing tool for microbial community analysis.

In NAND Flash-based SSDs, deduplication can provide an effective resolution of three critical issues: cell lifetime, write performance, and garbage collection overhead. However, deduplication at SSD device level distinguishes itself from the one at enterprise storage systems in many aspects, whose success lies in proper exploitation of underlying very limited hardware resources and workload characteristics of SSDs. In this paper, we develop a novel deduplication framework elaborately tailored for SSDs. We first mathematically develop an analytical model that enables us to calculate the minimum required duplication rate in order to achieve performance gain given deduplication overhead. Then, we explore a number of design choices for implementing deduplication components by hardware or software. As a result, we propose two acceleration techniques: sampling-based filtering and recency-based fingerprint management. The former selectively applies deduplication based upon sampling and the latter effectively exploits limited controller memory while maximizing the deduplication ratio. We prototype the proposed deduplication framework in three physical hardware platforms and investigate deduplication efficiency according to various CPU capabilities and hardware/software alternatives. Experimental results have shown that we achieve the duplication rate ranging from 4% to 51%, with an average of 17%, for the nine workloads considered in this work. The response time of a write request can be improved by up to 48% with an average of 15%, while the lifespan of SSDs is expected to increase up to 4.1 times with an average of 2.4 times.

Dynamic Voltage and Frequency Scaling (DVFS) is an effective low-power technique for real-time workloads. Its effectiveness critically depends on the accurate prediction of the task execution time. Many DVFS approaches have been proposed, but they are insufficient for highly nonstationary workloads. Several recent DVFS techniques adopted adaptive filters to improve accuracy. However, their improvement was rather limited, since they mainly focused on applying a filter framework to the target application without tuning it. We address this issue by proposing Particle Filter (PF)-based video decoders (MPEG2 and H.264) which exploit application-specific characteristics. More specifically, our PF-based video decoders utilize the size of each frame for the prediction of its decoding time. Compared to previous work, the PF is more suitable for our purpose, since it achieves higher prediction accuracy, even for highly nonstationary workloads such as H.264 clips. Our results show that the energy saved by the proposed approach is comparable to that of the ideal policy called oracle-DVFS, while the existing methods we tested were far inferior to oracle-DVFS in terms of H.264 video decoding. Additionally, when our method was used, only 0.40 and 6.88 percent of the frames missed their deadlines with negligible computational overhead for MPEG and H.264, respectively.

MOTIVATION: Capillary electrophoresis (CE) of nucleic acids is a workhorse technology underlying high-throughput genome analysis and large-scale chemical mapping for nucleic acid structural inference. Despite the wide availability of CE-based instruments, there remain challenges in leveraging their full power for quantitative analysis of RNA and DNA structure, thermodynamics and kinetics. In particular, the slow rate and poor automation of available analysis tools have bottlenecked a new generation of studies involving hundreds of CE profiles per experiment. RESULTS: We propose a computational method called high-throughput robust analysis for capillary electrophoresis (HiTRACE) to automate the key tasks in large-scale nucleic acid CE analysis, including the profile alignment that has heretofore been a rate-limiting step in the highest throughput experiments. We illustrate the application of HiTRACE on 13 datasets representing 4 different RNAs, 3 chemical modification strategies and up to 480 single mutant variants; the largest datasets each include 87 360 bands. By applying a series of robust dynamic programming algorithms, HiTRACE outperforms prior tools in terms of alignment and fitting quality, as assessed by measures including the correlation between quantified band intensities between replicate datasets. Furthermore, while the smallest of these datasets required 7-10 h of manual intervention using prior approaches, HiTRACE quantitation of even the largest datasets herein was achieved in 3-12 min. The HiTRACE method, therefore, resolves a critical barrier to the efficient and accurate analysis of nucleic acid structure in experiments involving tens of thousands of electrophoretic bands.

This paper presents a methodology for reducing interconnect resources in reconfigurable platforms such as field-programmable gate arrays (FPGAs). This methodology utilizes the techniques developed for the buffer-based dataflow, a new design representation suitable for implementing data-centric applications in a reconfigurable platform. In a buffer-based dataflow, nodes correspond to processing blocks and buffer controllers represent the interconnects between the processing blocks. Since we can isolate the functional execution and data transfer of each node by using buffer controllers, a buffer-based dataflow is helpful for reducing overall design time and for increasing reconfigurability. In this paper, we propose a sharing methodology that can reduce the buffer memory and the number of buses used in the realization of a buffer-based dataflow. By reducing the resources allocated to buffer controllers, we can achieve interconnect resource reduction. The proposed sharing methodology can increase the dynamic energy consumption due to the increased port-loading capacitance. By using the energy consumption model determined by the costs of buffers and buses, we investigate whether the sharing case with the minimum resources corresponds to the sharing case consuming the minimum energy or not. We evaluate the proposed sharing methodology with the dataflow graphs representing data-centric applications such as SIRF, IPv4, MC-CDMA transmitter and receiver.

This paper presents an effective design methodology which maps a complex system represented as a dataflow graph to a reconfigurable target architecture having multi-core processors and programmable logics. In order to synchronize data transfers between two processing blocks mapped to different processors (alternatively, one block is mapped to a processor and the other is realized as a hardware), we propose a mapping methodology that exploits the buffer-based dataflow, a new representation technique for realizing data-centric applications in reconfigurable platforms. From the buffer-based dataflow and estimated execution times of functional blocks and data transfers, the proposed methodology creates a mapped partition and generates the template code which runs on the processors of the target platform. We also use a processor initiation scheme to prevent wrong operations from happening when actual execution takes longer than estimated. Our proposed mapping methodology and the generated template code are evaluated with the SystemC model and Xilinx ISE.

On-chip interconnection network is a crucial design component in high-performance System-on-Chips (SoCs). Many of previous works have focused on the automation of its topology design, since the topology largely determines its overall performance. For this purpose, they mostly require a switch library which includes all possible switch configurations (e.g. the number of in/output ports and data width) with their implementation costs such as delay, area, and power. More precisely, they characterize the switches by synthesizing them with a common design objective (e.g. minimizing area) and common design constraints for a given gate-level design library. The implementation costs are used in evaluating the topologies throughout the topology synthesis. The major drawback of single switch library approach is that it forces the topology synthesis methods to search the best topology with the assumption that all the switches comprising a topology will be implemented (synthesized) with a common design objective and common design constraints. Such assumption prevents them from exploring diverse combinations of the switches for a topology from the implementation perspective. To tackle this issue, we propose a topology synthesis method with multiple switch libraries, where the switch libraries are prepared with different design objectives and design constraints. The experimental results show that the power consumption and the area of optimal topologies can be saved by up to 67.1% and 27.2%, respectively, by the proposed method with negligible synthesis time overhead.

As the cost per bit of NAND flash memory devices rapidly decreases, NAND-flash-based Solid-State Disks (SSDs) are replacing Hard Disk Drives (HDDs) used in a wide spectrum of consumer computing devices. Although typical SSDs can deliver higher performances than HDDs can, the full capabilities of SSDs are currently not exploited in most systems. This is because an SSD is interfaced with its host system using the architectures and interface protocols designed for HDDs, due to compatibility issues. Given the pace at which the stand-alone performance of SSDs improves, the performance loss of SSDs due to the legacy interface and system architecture will soon become intolerable. To address this issue, we propose several architectural choices to fully exploit the performance of SSDs used in consumer PC architectures. More specifically, we explore its interface scheme, and data transfer concurrency with the change of the conventional PC architecture if necessary. We evaluated the performance of the architectural choices by prototyping them with SystemC. The experimental results guide us how to trade off the performance enhancement and the change of the PC architecture. The performance improvement was maximized by 2.67 times when the PC architecture is changed to support a dual-port SSD connected to the North Bridge via the Double-Data Rate (DDR) interface in real trace environments.

In this letter, a numerically efficient algorithm is presented to compute the performance upper bounds for terminated convolutional codes based on a modified trellis diagram. A weight enumerator of one variable is defined with a new set of coefficients. The coefficients of the weight enumerator are shown to form a semiring.

MOTIVATION: For high-throughput prediction of the helical arrangements of large RNA molecules, an innovative method termed multiplexed hydroxyl radical (*OH) cleavage analysis (MOHCA) has been proposed. A key step in this promising technique is to detect peaks accurately from noisy radioactivity profiles. Since manual peak finding is laborious and prone to error, an automated peak detection method to improve the accuracy and throughput of MOHCA is required. Existing methods were not applicable to MOHCA due to their high false positive rates. RESULTS: We developed a two-step computational method that can detect peaks from MOHCA profiles in a robust manner. The first step exploits an ensemble of linear and non-linear signal processing techniques to find true peak candidates. In the second step, a binary classifier trained with the characteristics of true and false peaks is used to eliminate false peaks out of the peak candidates. We tested the proposed approach with 2002 MOHCA cleavage profiles and obtained the median recall, precision and F-measure values of 0.917, 0.750 and 0.830, respectively. Compared with the alternatives considered, the proposed method was able to handle false peaks substantially better, thus resulting in 51.0-71.8% higher median values of precision and F-measure. AVAILABILITY: The software and supplementary data are available at http://dna.korea.ac.kr/pub/mohca.

BACKGROUND: The problem of locating valid peaks from data corrupted by noise frequently arises while analyzing experimental data. In various biological and chemical data analysis tasks, peak detection thus constitutes a critical preprocessing step that greatly affects downstream analysis and eventual quality of experiments. Many existing techniques require the users to adjust parameters by trial and error, which is error-prone, time-consuming and often leads to incorrect analysis results. Worse, conventional approaches tend to report an excessive number of false alarms by finding fictitious peaks generated by mere noise. RESULTS: We have designed a novel peak detection method that can significantly reduce parameter sensitivity, yet providing excellent peak detection performance and negligible false alarm rates from gas chromatographic data. The key feature of our new algorithm is the successive use of peak enhancement algorithms that are deliberately designed for a gradual improvement of peak detection quality. We tested our approach with real gas chromatograms as well as intentionally contaminated spectra that contain Gaussian or speckle-type noise. CONCLUSION: Our results demonstrate that the proposed method can achieve near perfect peak detection performance while maintaining very small false alarm probabilities in case of gas chromatograms. Given the fact that biological signals appear in the form of peaks in various experimental data and that the propose method can easily be extended to such data, our approach will be a useful and robust tool that can help researchers highlight valid signals in their noisy measurements.

Dynamic voltage scaling (DVS) is a popular energy-saving technique for real-time tasks. The effectiveness of DVS critically depends on the accuracy of workload estimation, since DVS exploits the slack or the difference between the deadline and execution time. Many existing DVS techniques are profile based and simply utilize the worst-case or average execution time without estimation. Several recent approaches recognize the importance of workload estimation and adopt statistical estimation techniques. However, these approaches still require extensive profiling to extract reliable workload statistics and furthermore cannot effectively handle time-varying workloads. Feedback-control-based adaptive algorithms have been proposed to handle such nonstationary workloads, but their results are often too sensitive to parameter selection. To overcome these limitations of existing approaches, we propose a novel workload estimation technique for DVS. This technique is based on the Kalman filter and can estimate the processing time of workloads in a robust and accurate manner by adaptively calibrating estimation error by feedback. We tested the proposed method with workloads of various characteristics extracted from eight MPEG video clips. To thoroughly evaluate the performance of our approach, we used both a cycle-accurate simulator and an XScale-based test board. Our simulation result demonstrates that the proposed technique outperforms the compared alternatives with respect to the ability to meet given timing and Quality of Service constraints. Furthermore, we found that the accuracy of our approach is almost comparable to the oracle accuracy achievable only by offline analysis. Experimental results indicate that using our approach can reduce energy consumption by 57.5% on average, only with negligible deadline miss ratio (DMR) around 6.1%. Moreover, the average of computational overheads for the proposed technique is just 0.3%, which is the minimum value compared to other methods. More importantly, the DMR of our method is bounded by 11.7% in the worst case, while those of other methods are twice or more than ours.

The post-layout gate-level simulation constitutes a critical design step for timing closure. The major drawback of traditional post-layout gate-level simulation is its long analysis time, which becomes exacerbated as design complexity increases. An alternative method isstatictiminganalysis(STA), which can drastically reduce analysis time. However, STA sacrifices accuracy for speed and often produces unrealistic results such as false paths and overly pessimistic estimates. In this paper, we propose a hybrid analysis method that can significantly reduce analysis time, while preserving accuracy, with respect to the traditional gate-level simulation. Our key idea is that a large speedup would be possible by removing those events that are repetitious and unnecessary for simulation. In particular, we focus on reducing the number of clock-related events, which account for a major portion of all the events handled by a simulator. We tested the proposed method extensively with various benchmark circuits as well as industrial designs. Our experimental results exhibit that the proposed approach accelerates the total simulation speed by two times on average, yet maintaining the accuracy acquired by the traditional gate-level simulation.

BACKGROUND: Structural genomics initiatives are producing increasing numbers of three-dimensional (3D) structures for which there is little functional information. Structure-based annotation of molecular function is therefore becoming critical. We previously presented FEATURE, a method for describing microenvironments around functional sites in proteins. However, FEATURE uses supervised machine learning and so is limited to building models for sites of known importance and location. We hypothesized that there are a large number of sites in proteins that are associated with function that have not yet been recognized. Toward that end, we have developed a method for clustering protein microenvironments in order to evaluate the potential for discovering novel sites that have not been previously identified. RESULTS: We have prototyped a computational method for rapid clustering of millions of microenvironments in order to discover residues whose surrounding environments are similar and which may therefore share a functional or structural role. We clustered nearly 2,000,000 environments from 9,600 protein chains and defined 4,550 clusters. As a preliminary validation, we asked whether known 3D environments associated with PROSITE motifs were "rediscovered". We found examples of clusters highly enriched for residues that share PROSITE sequence motifs. CONCLUSION: Our results demonstrate that we can cluster protein environments successfully using a simplified representation and K-means clustering algorithm. The rediscovery of known 3D motifs allows us to calibrate the size and intercluster distances that characterize useful clusters. This information will then allow us to find new clusters with similar characteristics that represent novel structural or functional sites.

Co-clustering has not been much exploited in biomedical informatics, despite its success in other domains. Most of the previous applications were limited to analyzing gene expression data. We performed co-clustering analysis on other types of data and obtained promising results, as summarized in this paper.

In recent years, the development of high throughput devices for the massive parallel analyses of genomic data has lead to the generation of large amount of new biological evidences and has triggered the proliferation of data mining algorithms for the extraction of meaningful information. Microarrays for gene expression analyses are part of this revolution and provide important insight in molecular biology often in the form of coherent sets of genes representing previously uncharacterized processes. Large amount of data are continuously produced in this form, and computational approaches can significantly improve the efficient use of these results, since comparison among numbers of genes sets can give new meaningful information at no cost from the experimental biology point of view. To address this opportunity we designed and implemented FIT, a scalable, unsupervised algorithm that quantitatively compares different populations of gene sets using two distinct measures of similarity between any two gene sets. These measures are then used to obtain a summary statistic that describes the tightness of fit between sets belonging to two distinct populations of gene sets. We present the results of FIT on two data sets for the study of Lymphoma and Acute Lymphoblastic Leukemia. In both cases FIT was able to recapitulate the previous analyses on these datasets, to extend the results and to extract information likely to offer potential insights into the underlying biology.

Biochips are emerging as a useful tool for high-throughput acquisition of biological data and continue to grow in information quality and in discovering new applications. Recent advances include CMOS-based integrated biosensor arrays for deoxyribonucleic acid (DNA) expression analysis (Hassibi and Lee, 2005), (Schienle , 2004), and active research is ongoing for the miniaturization and integration of protein microarrays (Kiyonaka , 2004), (Rubina , 2003), (Scrivener , 2003), tissue microarrays (TMAs), (Chen , 2004), (Shergill , 2004), and fluorescence-based multiplexed cytokine immunoassays (Wang , 2002). The main advantages of microfluidic lab-on-a-chip include ease of use, speed of analysis, low sample and reagent consumption, and high reproducibility due to standardization and automation. Without effective data-analysis methods, however, the merit of acquiring massive data through biochips will be marginal. The high-dimensional nature of such data requires novel techniques that can cope with the curse of dimensionality better than conventional data-analysis approaches. In this paper, the authors proposed a pattern-mining method to analyze large-scale biological data obtained from high-throughput biochip experiments. In particular, when a data set is given as a matrix, the method can find patterns appearing in the form of (possibly overlapping) submatrices of the input matrix. The method exploits the techniques developed for the symbolic manipulation of Boolean functions. Leveraged by this approach, the method can find, given a data matrix, all patterns that satisfy specific input parameters. The authors tested the method with several large-scale biochip data and observed that the proposed method outperforms the alternatives in terms of efficiency and the number of patterns discovered.

The biclustering method can be a very useful analysis tool when some genes have multiple functions and experimental conditions are diverse in gene expression measurement. This is because the biclustering approach, in contrast to the conventional clustering techniques, focuses on finding a subset of the genes and a subset of the experimental conditions that together exhibit coherent behavior. However, the biclustering problem is inherently intractable, and it is often computationally costly to find biclusters with high levels of coherence. In this work, we propose a novel biclustering algorithm that exploits the zero-suppressedbinary decision diagrams (ZBDDs) data structure to cope with the computational challenges. Our method can find all biclusters that satisfy specific input conditions, and it is scalable to practical gene expression data. We also present experimental results confirming the effectiveness of our approach.

Clustering has been one of the most popular methods to discover useful biological insights from DNA microarray. An interesting paradigm is simultaneous clustering of both genes and experiments. This "biclustering" paradigm aims at discovering clusters that consist of a subset of the genes showing a coherent expression pattern over a subset of conditions. The pClustering approach is a technique that belongs to this paradigm. Despite many theoretical advantages, this technique has been rarely applied to actual gene expression data analysis. Possible reasons include the worst-case complexity of the clustering algorithm and the difficulty in interpreting clustering results. In this paper, we propose an enhanced framework for performing pClustering on actual gene expression analysis. Our new framework includes an effective data preparation method, highly scalable clustering strategies, and an intuitive result interpretation scheme. The experimental result confirms the effectiveness of our approach.