Unlike traditional classification tasks, recommendation is inherently subjective-whether an item should be suggested depends not only on user preferences and item semantics, but also on latent behavioral patterns and contextual cues. While recent LLM-based recommenders excel at modeling semantics and intent through generative reasoning, they often fail to capture collaborative signals and suffer from inefficiencies when applied globally across large interaction spaces. We propose Local Large Language Models for Recommendation(L3Rec), a novel model-agnostic framework that integrates collaborative filtering(CF) with generative LLMs through localized modeling. Our approach first applies a light-weight CF model to derive user and item embeddings, then clusters them into behaviorally coherent subgroups. Each cluster is assigned a dedicated generative LLM-referred to as a local LLM-trained only on its corresponding data subset. This enables fine-grained personalization while improving training efficiency through parallelism. At inference time, predictions from local models are aggregated via a fusion strategy, with a global CF fallback when needed. To the best of our knowledge, this is the first LLM-based recommendation framework to incorporate local collaborative structure. Experiments show that it achieves state-of-the-art performance with significantly better scalability and efficiency.

A variety of text-guided image editing models have been proposed recently. However, there is no widely-accepted standard evaluation method mainly due to the subjective nature of the task, letting researchers rely on manual user study. To address this, we introduce a novel Human-Aligned benchmark for Text-guided Image Editing (HATIE). Providing a large-scale benchmark set covering a wide range of editing tasks, it allows reliable evaluation, not limited to specific easy-to-evaluate cases. Also, HATIE provides a fully-automated and omnidirectional evaluation pipeline. Particularly, we combine multiple scores measuring various aspects of editing so as to align with human perception. We empirically verify that the evaluation of HATIE is indeed human-aligned in various aspects, and provide benchmark results on several state-of-the-art models to provide deeper insights on their performance.

Modern music retrieval systems often rely on fixed representations of user preferences, limiting their ability to capture users’ diverse and uncertain retrieval needs. To address this limitation, we introduce Diff4Steer, a novel generative retrieval framework that employs lightweight diffusion models to synthesize diverse seed embeddings representing potential directions for music exploration. Unlike deterministic methods that map user query to a single point in embedding space, Diff4Steer provides a statistical prior on the target modality (audio) for retrieval, effectively capturing the uncertainty and multi-faceted nature of user preferences. Furthermore, Diff4Steer can be steered by image or text inputs, enabling more flexible and controllable music discovery combined with nearest neighbor search. Our framework outperforms deterministic regression methods and LLM-based generative retrieval baseline in terms of retrieval and ranking metrics, demonstrating its effectiveness in capturing user preferences, leading to more diverse and relevant recommendations. Listening examples are available at tinyurl.com/diff4steer.

Video summarization is a task of shortening a video by choosing a subset of frames while preserving its essential moments. Despite the innate subjectivity of the task, previous works have deterministically regressed to an averaged frame score over multiple raters, ignoring the inherent subjectivity of what constitutes a good summary. We propose a novel problem formulation by framing video summarization as a conditional generation task, allowing a model to learn the distribution of good summaries and to generate multiple plausible summaries that better reflect varying human perspectives. Adopting diffusion models for the first time in video summarization, our proposed method, SummDiff, dynamically adapts to visual contexts and generates multiple candidate summaries conditioned on the input video. Extensive experiments demonstrate that SummDiff not only achieves the state-of-the-art performance on various benchmarks but also produces summaries that closely align with individual annotator preferences. Moreover, we provide a deeper insight with novel metrics from an analysis of the knapsack, which is an important last step of generating summaries but has been overlooked in evaluation.

In spite of the remarkable potential of Latent Diffusion Models (LDMs) in image generation, the desired properties and optimal design of the autoencoders have been underexplored. In this work, we analyze the role of autoencoders in LDMs and identify three key properties: latent smoothness, perceptual compression quality, and reconstruction quality. We demonstrate that existing autoencoders fail to simultaneously satisfy all three properties, and propose Variational Masked AutoEncoders (VMAEs), taking advantage of the hierarchical features maintained by Masked AutoEncoders. We integrate VMAEs into the LDM framework, introducing Latent Diffusion Models with Masked AutoEncoders (LDMAEs). Our code is available at https://github.com/isno0907/ldmae.

Graph convolutional networks have recently gained prominence in collaborative filtering (CF) for recommendations. However, we identify potential bottlenecks in two foundational components. First, the embedding layer leads to a latent space with limited capacity, overlooking locally observed but potentially valuable preference patterns. Also, the widely-used neighborhood aggregation is limited in its ability to leverage diverse preference patterns in a fine-grained manner. Building on spectral graph theory, we reveal that these limitations stem from graph filtering with a cut-off in the frequency spectrum and a restricted linear form. To address these issues, we introduce ChebyCF, a CF framework based on graph spectral filtering. Instead of a learned embedding, it takes a user's raw interaction history to utilize the full spectrum of signals contained in it. Also, it adopts Chebyshev interpolation to effectively approximate a flexible non-linear graph filter, and further enhances it by using an additional ideal pass filter and degree-based normalization. Through extensive experiments, we verify that ChebyCF overcomes the aforementioned bottlenecks and achieves state-of-the-art performance across multiple benchmarks and reasonably fast inference. Our code is available at https://github.com/snuviplab/ChebyCF.

Video-to-music generation demands both a temporally localized high-quality listening experience and globally aligned video-acoustic signatures. While recent music generation models excel at the former through advanced audio codecs, the exploration of video-acoustic signatures has been confined to specific visual scenarios. In contrast, our research confronts the challenge of learning globally aligned signatures between video and music directly from paired music and videos, without explicitly modeling domain-specific rhythmic or semantic relationships. We propose V2Meow, a video-to-music generation system capable of producing high-quality music audio for a diverse range of video input types using a multi-stage autoregressive model. Trained on 5k hours of music audio clips paired with video frames mined from in-the-wild music videos, V2Meow is competitive with previous domain-specific models when evaluated in a zero-shot manner. It synthesizes high-fidelity music audio waveforms solely by conditioning on pre-trained general-purpose visual features extracted from video frames, with optional style control via text prompts. Through both qualitative and quantitative evaluations, we demonstrate that our model outperforms various existing music generation systems in terms of visual-audio correspondence and audio quality. Music samples are available at tinyurl.com/v2meow.

In this paper, we propose and conduct a comprehensive benchmark on moment localization task, which aims to retrieve a segment that corresponds to a text query from a single untrimmed video. Our study starts from an observation that most moment localization papers report experimental results only on a few datasets in spite of availability of far more benchmarks. Thus, we conduct an extensive benchmark study to measure the performance of representative methods on widely used 7 datasets. Looking further into the details, we pose additional research questions and empirically verify them, including if they rely on unintended biases introduced by specific training data, if advanced visual features trained on classification task transfer well to this task, and if computational cost of each model pays off. With a series of these experiments, we provide multi-faceted evaluation of state-of-the-art moment localization models. Codes are available at \url{https://github.com/snuviplab/MoLEF}.

While backpropagation (BP) has achieved widespread success in deep learning, it faces two prominent challenges: computational inefficiency and biological implausibility. In response to these challenges, local supervision, encompassing Local Learning (LL) and Forward Learning (FL), has emerged as a promising research direction. LL employs module-wise BP to achieve competitive results yet relies on module-wise auxiliary networks, which increase memory and parameter demands. Conversely, FL updates layer weights without BP and auxiliary networks but falls short of BP’s performance. This paper proposes a simple yet effective objective within a contrastive learning framework for local supervision without auxiliary networks. Given the insight that the existing contrastive learning framework for local supervision is susceptible to task-irrelevant information without auxiliary networks, we present DICTIONARY CONTRASTIVE LEARNING (DCL) that optimizes the similarity between local features and label embeddings. Our method using static label embeddings yields substantial performance improvements in the FL scenario, outperforming state-of-the-art FL approaches. Moreover, our method using adaptive label embeddings closely approaches the performance achieved by LL while achieving superior memory and parameter efficiency.

The latent space of diffusion model mostly still remains unexplored, despite its great success and potential in the field of generative modeling. In fact, the latent space of existing diffusion models are entangled, with a distorted mapping from its latent space to image space. To tackle this problem, we present Isometric Diffusion, equipping a diffusion model with a geometric regularizer to guide the model to learn a geometrically sound latent space of the training data manifold. This approach allows diffusion models to learn a more disentangled latent space, which enables smoother interpolation, more accurate inversion, and more precise control over attributes directly in the latent space. Our extensive experiments consisting of image interpolations, image inversions, and linear editing show the effectiveness of our method.

Masked Autoencoder (MAE) is a self-supervised approach for representation learning, widely applicable to a variety of downstream tasks in computer vision. In spite of its success, it is still not fully uncovered what and how MAE exactly learns. In this paper, with an in-depth analysis, we discover that MAE intrinsically learns pattern-based patch-level clustering from surprisingly early stages of pre-training. Upon this understanding, we propose self-guided masked autoencoder, which internally generates informed mask by utilizing its progress in patch clustering, substituting the naive random masking of the vanilla MAE. Our approach significantly boosts its learning process without relying on any external models or supplementary information, keeping the benefit of self-supervised nature of MAE intact. Comprehensive experiments on various downstream tasks verify the effectiveness of the proposed method.

Graph convolutions have been successfully applied to recommendation systems, utilizing high-order collaborative signals present in the user-item interaction graph. This idea, however, has not been applicable to the cold-start items, since cold nodes are isolated in the graph and thus do not take advantage of information exchange from neighboring nodes. Recently, there have been a few attempts to utilize graph convolutions on item-item or user-user attribute graphs to capture high-order collaborative signals for cold-start cases, but these approaches are still limited in that the item-item or user-user graph falls short in capturing the dynamics of user-item interactions, as their edges are constructed based on arbitrary and heuristic attribute similarity.

Facial expression recognition (FER) has greatly benefited from deep learning but still faces challenges in dataset collection due to the nuanced nature of facial expressions. In this study, we present a novel unlabeled dataset and semi-supervised contrastive learning framework that utilizes Reaction Mashup (RM) videos, a video that includes multiple individuals reacting to the same film. We created a Reaction Mashup dataset (RMset) from these videos. Our framework integrates three distinct modules: A classification module for supervised facial expression categorization, an attention module for inter-sample attention learning, and a contrastive module for attention-based contrastive learning using RMset. We utilize both the classification and attention modules for the initial training, subsequently incorporating the contrastive module to enhance the learning process. Our experiments demonstrate that our method improves feature learning and outperforms state-of-the-art models on three benchmark FER datasets. Codes are available at https://github.com/yunseongcho/RMFER.

Novel view synthesis has been heavily driven by NeRF-based models, but these models often hold limitations with the requirement of dense coverage of input views and expensive computations. NeRF models designed for scenarios with a few sparse input views face difficulty in being generalizable to complex or unbounded scenes, where multiple scene content can be at any distance from a multi-directional camera, and thus generate unnatural and low quality images with blurry or floating artifacts. To accommodate the lack of dense information in sparse view scenarios and the computational burden of NeRF-based models in novel view synthesis, our approach adopts diffusion models. In this paper, we present PoseDiff, which combines the fast and plausible generation ability of diffusion models and 3D-aware view consistency of pose parameters from NeRF-based models. Specifically, PoseDiff is a multimodal pose-conditioned diffusion model applicable for novel view synthesis of unbounded scenes as well as bounded or forward-facing scenes with sparse views. PoseDiff renders plausible novel views for given pose parameters while maintaining high-frequency geometric details in significantly less time than conventional NeRF-based methods.

Zero-shot learning offers an efficient solution for a machine learning model to treat unseen categories, avoiding exhaustive data collection. Zero-shot Sketch-based Image Retrieval (ZS-SBIR) simulates real-world scenarios where it is hard and costly to collect paired sketch-photo samples. We propose a novel framework that indirectly aligns sketches and photos by contrasting them through texts, removing the necessity of access to sketch-photo pairs. With an explicit modality encoding learned from data, our approach disentangles modality-agnostic semantics from modality-specific information, bridging the modality gap and enabling effective cross-modal content retrieval within a joint latent space. From comprehensive experiments, we verify the efficacy of the proposed model on ZS-SBIR, and it can be also applied to generalized and fine-grained settings.

X-ray computed tomography (CT) is one of the most common imaging techniques used to diagnose various diseases in the medical field. Its high contrast sensitivity and spatial resolution allow the physician to observe details of body parts such as bones, soft tissue, blood vessels, etc. As it involves potentially harmful radiation exposure to patients and surgeons, however, reconstructing 3D CT volume from perpendicular 2D X-ray images is considered a promising alternative, thanks to its lower radiation risk and better accessibility. This is highly challenging though, since it requires reconstruction of 3D anatomical information from 2D images with limited views, where all the information is overlapped. In this paper, we propose PerX2CT, a novel CT reconstruction framework from X-ray that reflects the perspective projection scheme. Our proposed method provides a different combination of features for each coordinate which implicitly allows the model to obtain information about the 3D location. We reveal the potential to reconstruct the selected part of CT with high resolution by properly using the coordinate-wise local and global features. Our approach shows potential for use in clinical applications with low computational complexity and fast inference time, demonstrating superior performance than baselines in multiple evaluation metrics.

This is an exploratory study that discovers the current image quantization (vector quantization) do not satisfy translation equivariance in the quantized space due to aliasing. Instead of focusing on anti-aliasing, we propose a simple yet effective way to achieve translation-equivariant image quantization by enforcing orthogonality among the codebook embeddings. To explore the advantages of translation-equivariant image quantization, we conduct three proof-of-concept experiments with a carefully controlled dataset: (1) text-to-image generation, where the quantized image indices are the target to predict, (2) image-to-text generation, where the quantized image indices are given as a condition, (3) using a smaller training set to analyze sample efficiency. From the strictly controlled experiments, we empirically verify that the translation-equivariant image quantizer improves not only sample efficiency but also the accuracy over VQGAN up to +11.9% in text-to-image generation and +3.9% in image-to-text generation.

Recent studies have substantiated that machine learning algorithms including convolutional neural networks often suffer from unreliable generalizations when there is a significant gap between the source and target data distributions. To mitigate this issue, a predetermined distribution shift has been addressed independently (e.g., single domain generalization, de-biasing). However, a distribution mismatch cannot be clearly estimated because the target distribution is unknown at training. Therefore, a conservative approach robust on unexpected diverse distributions is more desirable in practice. Our work starts from a motivation to allow adaptive inference once we know the target, since it is accessible only at testing. Instead of assuming and fixing the target distribution at training, our proposed approach allows adjusting the feature space the model refers to at every prediction, i.e., instance-wise adaptive inference. The extensive evaluation demonstrates our method is effective for generalization on diverse distributions.

3D pose estimation is an invaluable task in computer vision with various practical applications. Especially, 3D pose estimation for multi-person from a monocular video (3DMPPE) is particularly challenging and is still largely uncharted, far from applying to in-the-wild scenarios yet. We pose three unresolved issues with the existing methods: lack of robustness on unseen views during training, vulnerability to occlusion, and severe jittering in the output. As a remedy, we propose POTR-3D, the first realization of a sequence-to-sequence 2D-to-3D lifting model for 3DMPPE, powered by a novel geometry-aware data augmentation strategy, capable of generating unbounded data with a variety of views while caring about the ground plane and occlusions. Through extensive experiments, we verify that the proposed model and data augmentation robustly generalizes to diverse unseen views, robustly recovers the poses against heavy occlusions, and reliably generates more natural and smoother outputs. The effectiveness of our approach is verified not only by achieving the state-of-the-art performance on public benchmarks, but also by qualitative results on more challenging in-the-wild videos. Demo videos are available at https://www.youtube.com/@potr3d.

Sequence prediction on temporal data requires the ability to understand compositional structures of multi-level semantics beyond individual and contextual properties of parts. The task of temporal action segmentation remains challenging for the reason, aiming at translating an untrimmed activity video into a sequence of action segments. This paper addresses the problem by introducing an effective activity grammar to guide neural predictions for temporal action segmentation. We propose a novel grammar induction algorithm, dubbed KARI, that extracts a powerful context-free grammar from action sequence data. We also develop an efficient generalized parser, dubbed BEP, that transforms frame-level probability distributions into a reliable sequence of actions according to the induced grammar with recursive rules. Our approach can be combined with any neural network for temporal action segmentation to enhance the sequence prediction and discover its compositional structure. Experimental results demonstrate that our method significantly improves temporal action segmentation in terms of both performance and interpretability on two standard benchmarks, Breakfast and 50 Salads.

AI alignment refers to models acting towards human-intended goals, preferences, or ethical principles. Analyzing the similarity between models and humans can be a proxy measure for ensuring AI safety. In this paper, we focus on the models' visual perception alignment with humans, further referred to as AI-human visual alignment. Specifically, we propose a new dataset for measuring AI-human visual alignment in terms of image classification. In order to evaluate AI-human visual alignment, a dataset should encompass samples with various scenarios and have gold human perception labels. Our dataset consists of three groups of samples, namely Must-Act (i.e., Must-Classify), Must-Abstain, and Uncertain, based on the quantity and clarity of visual information in an image and further divided into eight categories. All samples have a gold human perception label; even Uncertain (e.g., severely blurry) sample labels were obtained via crowd-sourcing. The validity of our dataset is verified by sampling theory, statistical theories related to survey design, and experts in the related fields. Using our dataset, we analyze the visual alignment and reliability of five popular visual perception models and seven abstention methods. Our code and data is available at https://github.com/jiyounglee-0523/VisAlign.

Video highlight detection is a task to automatically select the most engaging moments from a long video. This problem is highly challenging since it aims to learn a general way of finding highlights from a variety of videos in the real world.The task has an innate subjectivity because the definition of a highlight differs across individuals. Therefore, to detect consistent and meaningful highlights, prior benchmark datasets have been labeled by multiple (5-20) raters. Due to the high cost of manual labeling, most existing public benchmarks are in extremely small scale, containing only a few tens or hundreds of videos. This insufficient benchmark scale causes multiple issues such as unstable evaluation or high sensitivity in traintest splits. We present Mr. HiSum, a large-scale dataset for video highlight detection and summarization, containing 31,892 videos and reliable labels aggregated over 50,000+ users per video. We empirically prove reliability of the labels as frame importance by cross-dataset transfer and user study.

Estimating the energetic properties of molecular systems is a critical task in material design. Machine learning has shown remarkable promise on this task over classical force fields, but a fully data-driven approach suffers from limited labeled data; not just the amount of available data lacks, but the distribution of labeled examples is highly skewed to stable states. In this work, we propose a molecular representation learning method that extrapolates well beyond the training distribution, powered by physics-driven parameter estimation from classical energy equations and self-supervised learning inspired from masked language modeling. To ensure reliability of the proposed model, we introduce a series of novel evaluation schemes in multifaceted ways, beyond the energy or force accuracy that has been dominantly used. From extensive experiments, we demonstrate that the proposed method is effective in discovering molecular structures, outperforming other baselines. Furthermore, we extrapolate it to the chemical reaction pathways beyond stable states, taking a step towards physically reliable molecular representation learning.

The task of predicting future actions from a video is crucial for a real-world agent interacting with others. When anticipating actions in the distant future, we humans typically consider long-term relations over the whole sequence of actions, i.e., not only observed actions in the past but also potential actions in the future. In a similar spirit, we propose an end-to-end attention model for action anticipation, dubbed Future Transformer (FUTR), that leverages global attention over all input frames and output tokens to predict a minutes-long sequence of future actions. Unlike the previous autoregressive models, the proposed method learns to predict the whole sequence of future actions in parallel decoding, enabling more accurate and fast inference for long-term anticipation. We evaluate our method on two standard benchmarks for long-term action anticipation, Breakfast and 50 Salads, achieving state-of-the-art results.

Although convolutional neural networks (CNNs) achieve state-of-the-art in image classification, recent works address their unreliable predictions due to their excessive dependence on biased training data. Existing unbiased modeling postulates that the bias in the dataset is obvious to know, but it is actually unsuited for image datasets including countless sensory attributes. To mitigate this issue, we present a new scenario that does not necessitate a predefined bias. Under the observation that CNNs do have multi-variant and unbiased representations in the model, we propose a conservative framework that employs this internal information for unbiased learning. Specifically, this mechanism is implemented via hierarchical features captured along the multiple layers and orthogonal regularization. Extensive evaluations on public benchmarks demonstrate our method is effective for unbiased learning.11Source code: https://github.com/aandyjeon/UBNet

The recent advances in representation learning inspire us to take on the challenging problem of unsupervised image classification tasks in a principled way. We propose ContraCluster, an unsupervised image classification method that combines clustering with the power of contrastive self-supervised learning. ContraCluster consists of three stages: (1) contrastive self-supervised pre-training (CPT), (2) contrastive prototype sampling (CPS), and (3) prototype-based semi-supervised fine-tuning (PB-SFT). CPS can select highly accurate, categorically prototypical images in an embedding space learned by contrastive learning. We use sampled prototypes as noisy labeled data to perform semi-supervised fine-tuning (PB-SFT), leveraging small prototypes and large unlabeled data to further enhance the accuracy. We demonstrate empirically that ContraCluster achieves new state-of-the-art results for standard benchmark datasets including CIFAR-10, STL-10, and ImageNet-10. For example, ContraCluster achieves about 90.8% accuracy for CIFAR-10, which outperforms DAC (52.2%), IIC (61.7%), and SCAN (87.6%) by a large margin. Without any labels, ContraCluster can achieve a 90.8% accuracy that is comparable to 95.8% by the best supervised counterpart.

We propose a self-supervised learning method for long text documents based on contrastive learning. A key to our method is Shuffle and Divide (SaD), a simple text augmentation algorithm that sets up a pretext task required for contrastive updates to BERT-based document embedding. SaD splits a document into two sub-documents containing randomly shuffled words in the entire documents. The sub-documents are considered positive examples, leaving all other documents in the corpus as negatives. After SaD, we repeat the contrastive update and clustering phases until convergence. It is naturally a time-consuming, cumbersome task to label text documents, and our method can help alleviate human efforts, which are most expensive resources in AI. We have empirically evaluated our method by performing unsupervised text classification on the 20 Newsgroups, Reuters-21578, BBC, and BBCSport datasets. In particular, our method pushes the current state-of-the-art, SS-SB-MT, on 20 Newsgroups by 20.94% in accuracy. We also achieve the state-of-the-art performance on Reuters-21578 and exceptionally-high accuracy performances (over 95%) for unsupervised classification on the BBC and BBCSport datasets.

Implicit feedback has been widely used to build commercial recommender systems. Because observed feedback represents users' click logs, there is a semantic gap between true relevance and observed feedback. More importantly, observed feedback is usually biased towards popular items, thereby overestimating the actual relevance of popular items. Although existing studies have developed unbiased learning methods using inverse propensity weighting (IPW) or causal reasoning, they solely focus on eliminating the popularity bias of items. In this paper, we propose a novel unbiased recommender learning model, namely BIlateral SElf-unbiased Recommender (BISER), to eliminate the exposure bias of items caused by recommender models. Specifically, BISER consists of two key components: (i) self-inverse propensity weighting (SIPW) to gradually mitigate the bias of items without incurring high computational costs; and (ii) bilateral unbiased learning (BU) to bridge the gap between two complementary models in model predictions, i.e., user- and item-based autoencoders, alleviating the high variance of SIPW. Extensive experiments show that BISER consistently outperforms state-of-the-art unbiased recommender models over several datasets, including Coat, Yahoo! R3, MovieLens, and CiteULike.

Session-based recommendation (SR) predicts the next items from a sequence of previous items consumed by an anonymous user. Most existing SR models focus only on modeling intra-session characteristics but pay less attention to inter-session relationships of items, which has the potential to improve accuracy. Another critical aspect of recommender systems is computational efficiency and scalability, considering practical feasibility in commercial applications. To account for both accuracy and scalability, we propose a novel session-based recommendation with a random walk, namely S-Walk. Precisely, S-Walk effectively captures intra- and inter-session correlations by handling high-order relationships among items using random walks with restart (RWR). By adopting linear models with closed-form solutions for transition and teleportation matrices that constitute RWR, S-Walk is highly efficient and scalable. Extensive experiments demonstrate that S-Walk achieves comparable or state-of-the-art performance in various metrics on four benchmark datasets. Moreover, the model learned by S-Walk can be highly compressed without sacrificing accuracy, conducting two or more orders of magnitude faster inference than existing DNN-based models, making it suitable for large-scale commercial systems.

Video generation models often operate under the assumption of fixed frame rates, which leads to suboptimal performance when it comes to handling flexible frame rates (e.g., increasing the frame rate of the more dynamic portion of the video as well as handling missing video frames). To resolve the restricted nature of existing video generation models' ability to handle arbitrary timesteps, we propose continuous-time video generation by combining neural ODE (Vid-ODE) with pixel-level video processing techniques. Using ODE-ConvGRU as an encoder, a convolutional version of the recently proposed neural ODE, which enables us to learn continuous-time dynamics, Vid-ODE can learn the spatio-temporal dynamics of input videos of flexible frame rates. The decoder integrates the learned dynamics function to synthesize video frames at any given timesteps, where the pixel-level composition technique is used to maintain the sharpness of individual frames. With extensive experiments on four real-world video datasets, we verify that the proposed Vid-ODE outperforms state-of-the-art approaches under various video generation settings, both within the trained time range (interpolation) and beyond the range (extrapolation). To the best of our knowledge, Vid-ODE is the first work successfully performing continuous-time video generation using real-world videos.

This work presents a generalized local factor model, namely Local Collaborative Autoencoders (LOCA). To our knowledge, it is the first generalized framework under the local low-rank assumption that builds on the neural recommendation models. We explore a large number of local models by adopting a generalized framework with different weight schemes for training and aggregating them. Besides, we develop a novel method of discovering a sub-community to maximize the coverage of local models. Our experimental results demonstrate that LOCA is highly scalable, achieving state-of-the-art results by outperforming existing AE-based and local latent factor models on several large-scale public benchmarks.

Session-based recommendation aims at predicting the next item given a sequence of previous items consumed in the session, e.g., on e-commerce or multimedia streaming services. Specifically, session data exhibits some unique characteristics, i.e., session consistency and sequential dependency over items within the session, repeated item consumption, and session timeliness. In this paper, we propose simple-yet-effective linear models for considering the holistic aspects of the sessions. The comprehensive nature of our models helps improve the quality of session-based recommendation. More importantly, it provides a generalized framework for reflecting different perspectives of session data. Furthermore, since our models can be solved by closed-form solutions, they are highly scalable. Experimental results demonstrate that the proposed linear models show competitive or state-of-the-art performance in various metrics on several real-world datasets.

Although essential to revealing biased performance, well intentioned attempts at algorithmic auditing can have effects that may harm the very populations these measures are meant to protect. This concern is even more salient while auditing biometric systems such as facial recognition, where the data is sensitive and the technology is often used in ethically questionable manners. We demonstrate a set of fiveethical concerns in the particular case of auditing commercial facial processing technology, highlighting additional design considerations and ethical tensions the auditor needs to be aware of so as not exacerbate or complement the harms propagated by the audited system. We go further to provide tangible illustrations of these concerns, and conclude by reflecting on what these concerns mean for the role of the algorithmic audit and the fundamental product limitations they reveal.

Representation learning is widely applied for various tasks on multimedia data, e.g., retrieval and search. One approach for learning useful representation is by utilizing the relationships or similarities between examples. In this work, we explore two promising scalable representation learning approaches on video domain. With hierarchical graph clusters built upon video-to-video similarities, we propose: 1) smart negative sampling strategy that significantly boosts training efficiency with triplet loss, and 2) a pseudo-classification approach using the clusters as pseudo-labels. The embeddings trained with the proposed methods are competitive on multiple video understanding tasks, including related video retrieval and video annotation. Both of these proposed methods are highly scalable, as verified by experiments on large-scale datasets.

Glaucoma is one of the leading causes of irreversible but preventable blindness in working age populations. Color fundus photography (CFP) is the most cost-effective imaging modality to screen for retinal disorders. However, its application to glaucoma has been limited to the computation of a few related biomarkers such as the vertical cup-to-disc ratio. Deep learning approaches, although widely applied for medical image analysis, have not been extensively used for glaucoma assessment due to the limited size of the available data sets. Furthermore, the lack of a standardize benchmark strategy makes difficult to compare existing methods in a uniform way. In order to overcome these issues we set up the Retinal Fundus Glaucoma Challenge, REFUGE (https://refuge.grand-challenge.org), held in conjunction with MICCAI 2018. The challenge consisted of two primary tasks, namely optic disc/cup segmentation and glaucoma classification. As part of REFUGE, we have publicly released a data set of 1200 fundus images with ground truth segmentations and clinical glaucoma labels, currently the largest existing one. We have also built an evaluation framework to ease and ensure fairness in the comparison of different models, encouraging the development of novel techniques in the field. 12 teams qualified and participated in the online challenge. This paper summarizes their methods and analyzes their corresponding results. In particular, we observed that two of the top-ranked teams outperformed two human experts in the glaucoma classification task. Furthermore, the segmentation results were in general consistent with the ground truth annotations, with complementary outcomes that can be further exploited by ensembling the results.

Video is one of the richest sources of information available online but extracting deep insights from video content at internet scale is still an open problem, both in terms of depth and breadth of understanding, as well as scale. Over the last few years, the field of video understanding has made great strides due to the availability of large-scale video datasets and core advances in image, audio, and video modeling architectures. However, the state-of-the-art architectures on small scale datasets are frequently impractical to deploy at internet scale, both in terms of the ability to train such deep networks on hundreds of millions of videos, and to deploy them for inference on billions of videos. In this paper, we present a MapReduce-based training framework, which exploits both data parallelism and model parallelism to scale training of complex video models. The proposed framework uses alternating optimization and full-batch fine-tuning, and supports large Mixture-of-Experts classifiers with hundreds of thousands of mixtures, which enables a trade-off between model depth and breadth, and the ability to shift model capacity between shared (generalization) layers and per-class (specialization) layers. We demonstrate that the proposed framework is able to reach state-of-the-art performance on the largest public video datasets, YouTube-8M and Sports-1M, and can scale to 100 times larger datasets.

Tracking and predicting extreme events in large-scale spatio-temporal climate data are long standing challenges in climate science. In this paper, we propose Convolutional LSTM (ConvLSTM)-based spatio-temporal models to track and predict hurricane trajectories from large-scale climate data; namely, pixel-level spatio-temporal history of tropical cyclones. To address the tracking problem, we model time-sequential density maps of hurricane trajectories, enabling to capture not only the temporal dynamics but also spatial distribution of the trajectories. Furthermore, we introduce a new trajectory prediction approach as a problem of sequential forecasting from past to future hurricane density map sequences. Extensive experiment on actual 20 years record shows that our ConvLSTM-based tracking model significantly outperforms existing approaches, and that the proposed forecasting model achieves successful mapping from predicted density map to ground truth.

The goal of video understanding is to develop algorithms that enable machines understand videos at the level of human experts. Researchers have tackled various domains including video classification, search, personalized recommendation, and more. However, there is a research gap in combining these domains in one unified learning framework. Towards that, we propose a deep network that embeds videos using their audio-visual content, onto a metric space which preserves video-to-video relationships. Then, we use the trained embedding network to tackle various domains including video classification and recommendation, showing significant improvements over state-of-the-art baselines. The proposed approach is highly scalable to deploy on large-scale video sharing platforms like YouTube.

Nonnegative matrix factorization (NMF) has been increasingly popular for topic modeling of large-scale documents. However, the resulting topics often represent only general, thus redundant information about the data rather than minor, but potentially meaningful information to users. To tackle this problem, we propose a novel ensemble model of nonnegative matrix factorization for discovering high-quality local topics. Our method leverages the idea of an ensemble model to successively perform NMF given a residual matrix obtained from previous stages and generates a sequence of topic sets. The novelty of our method lies in the fact that it utilizes the residual matrix inspired by a state-of-the-art gradient boosting model and applies a sophisticated local weighting scheme on the given matrix to enhance the locality of topics, which in turn delivers high-quality, focused topics of interest to users.

Nonnegative matrix factorization (NMF) has been widely applied in many domains. In document analysis, it has been increasingly used in topic modeling applications, where a set of underlying topics are revealed by a low-rank factor matrix from NMF. However, it is often the case that the resulting topics give only general topic information in the data, which tends not to convey much information. To tackle this problem, we propose a novel ensemble model of nonnegative matrix factorization for discovering high-quality local topics. Our method leverages the idea of an ensemble model, which has been successful in supervised learning, into an unsupervised topic modeling context. That is, our model successively performs NMF given a residual matrix obtained from previous stages and generates a sequence of topic sets. Our algorithm for updating the input matrix has novelty in two aspects. The first lies in utilizing the residual matrix inspired by a state-of-the-art gradient boosting model, and the second stems from applying a sophisticated local weighting scheme on the given matrix to enhance the locality of topics, which in turn delivers high-quality, focused topics of interest to users. We evaluate our proposed method by comparing it against other topic modeling methods, such as a few variants of NMF and latent Dirichlet allocation, in terms of various evaluation measures representing topic coherence, diversity, coverage, computing time, and so on. We also present qualitative evaluation on the topics discovered by our method using several real-world data sets.

Matrix approximation is a common tool in recommendation systems, text mining, and computer vision. A prevalent assumption in constructing matrix approximations is that the partially observed matrix is low-rank. In this paper, we propose, analyze, and experiment with two procedures, one parallel and the other global, for constructing local matrix approximations. The two approaches approximate the observed matrix as a weighted sum of low-rank matrices. These matrices are limited to a local region of the observed matrix. We analyze the accuracy of the proposed local low-rank modeling. Our experiments show improvements in prediction accuracy over classical approaches for recommendation tasks.

Sentiment analysis predicts a one-dimensional quantity describing the positive or negative emotion of an author. Mood analysis extends the one-dimensional sentiment response to a multi-dimensional quantity, describing a diverse set of human emotions. In this paper, we extend sentiment and mood analysis temporally and model emotions as a function of time based on temporal streams of blog posts authored by a specific author. The model is useful for constructing predictive models and discovering scientific models of human emotions.

The n-gram model has been widely used to capture the local ordering of words, yet its exploding feature space often causes an estimation issue. This paper presents local context sparse coding (LCSC), a non-probabilistic topic model that effectively handles large feature spaces using sparse coding. In addition, it introduces a new concept of locality, local contexts, which provides a representation that can generate locally coherent topics and document representations. Our model efficiently finds topics and representations by applying greedy coordinate descent updates. The model is useful for discovering local topics and the semantic flow of a document, as well as constructing predictive models.

Users today are constantly switching back and forth from applications where they consume or create content (such as e-books and productivity suites like Microsoft Office and Google Docs) to search engines where they satisfy their information needs. Unfortunately, though, this leads to a suboptimal user experience as the search engine lacks any knowledge about the content that the user is authoring or consuming in the application. As a result, productivity suites are starting to incorporate features that let the user "explore while they work". Existing work in the literature that can be applied to this problem takes a standard bag-of-words information retrieval approach, which consists of automatically creating a query that includes not only the target phrase or entity chosen by the user but also relevant terms from the context. While these approaches have been successful, they are inherently limited to returning results (documents) that have a syntactic match with the keywords in the query.

We propose an efficient low-overhead methodology for screening key layouts for ultimate typing speed. It is fast and complementary to existing protocols in other ways. For equal overhead, it allows testing over a wider range of users. It is subject to potential biases, but they can be quantified and adjusted. It assumes an existing standard layout which is used as a comparator. It eliminates bias from differing familiarity and finger memory by appropriately mapping both layouts. We illustrate it with two mobile phone keypad layouts: Samsung's variant of the common ABC layout, and a new user-specific layout (PM). We repeat a comparison previously undertaken by a training-based method, which used ten participants, and estimated a $54\%$ speedup for PM (SD = $35\%$). The new method used 116 participants and estimated a $19.5\%$ speedup (SD = $7.5\%$). Differences in speedup estimates can be explained by the wide confidence interval of the training-based methods, differences in test corpuses, and the inherent conservatism of the new method. Effects of user characteristics could be meaningfully tested due to the larger test group. Gender had no detectable effect on any of the measures, but age and keypad experience were significantly related to ABC and PM performance. However, the relative speedup was unaffected by age or keypad experience: the method can remove comparison biases arising from differing experience levels.

Personalized recommendation systems are used in a wide variety of applications such as electronic commerce, social networks, web search, and more. Collaborative filtering approaches to recommendation systems typically assume that the rating matrix (e.g., movie ratings by viewers) is low-rank. In this paper, we examine an alternative approach in which the rating matrix is locally low-rank. Concretely, we assume that the rating matrix is low-rank within certain neighborhoods of the metric space defined by (user, item) pairs. We combine a recent approach for local low-rank approximation based on the Frobenius norm with a general empirical risk minimization for ranking losses. Our experiments indicate that the combination of a mixture of local low-rank matrices each of which was trained to minimize a ranking loss outperforms many of the currently used state-of-the-art recommendation systems. Moreover, our method is easy to parallelize, making it a viable approach for large scale real-world rank-based recommendation systems.

Matrix approximation is a common tool in recommendation systems, text mining, and computer vision. A prevalent assumption in constructing matrix approximations is that the partially observed matrix is of low-rank. We propose a new matrix approximation model where we assume instead that the matrix is locally of low-rank, leading to a representation of the observed matrix as a weighted sum of low-rank matrices. We analyze the accuracy of the proposed local low-rank modeling. Our experiments show improvements in prediction accuracy over classical approaches for recommendation tasks.

For cold-start recommendation, it is important to rapidly profile new users and generate a good initial set of recommendations through an interview process --- users should be queried adaptively in a sequential fashion, and multiple items should be offered for opinion solicitation at each trial. In this work, we propose a novel algorithm that learns to conduct the interview process guided by a decision tree with multiple questions at each split. The splits, represented as sparse weight vectors, are learned through an L_1-constrained optimization framework. The users are directed to child nodes according to the inner product of their responses and the corresponding weight vector. More importantly, to account for the variety of responses coming to a node, a linear regressor is learned within each node using all the previously obtained answers as input to predict item ratings. A user study, preliminary but first in its kind in cold-start recommendation, is conducted to explore the efficient number and format of questions being asked in a recommendation survey to minimize user cognitive efforts. Quantitative experimental validations also show that the proposed algorithm outperforms state-of-the-art approaches in terms of both the prediction accuracy and user cognitive efforts.

Recent approaches to collaborative filtering have concentrated on estimating an algebraic or statistical model, and using the model for predicting missing ratings. In this paper we observe that different models have relative advantages in different regions of the input space. This motivates our approach of using stagewise linear combinations of collaborative filtering algorithms, with non-constant combination coefficients based on kernel smoothing. The resulting stagewise model is computationally scalable and outperforms a wide selection of state-of-the-art collaborative filtering algorithms.

We present a rapid screening method for keyboard layouts. The method relies on user familiarity with a pre-existing layout, and can provide rapid, low-cost estimates of the eventual user speed on the new layout. It provides an effective pre-screening for the expensive training curve estimates of previous methods, and provides valuable supplementary information not available from them. We provide preliminary validation of the method on a mobile phone layout problem.