Recently, 3D Gaussian Splatting (3D-GS) has emerged, showing real-time rendering speeds and high-quality results in static scenes. Although several 3D-GS methods show effectiveness in static scenes, their performance significantly degrades in real-world environments due to transient objects, lighting variations, and diverse levels of occlusion. To tackle this, existing methods estimate occluders or transient elements by leveraging pre-trained models or integrating additional transient field pipelines. However, these methods still suffer from two defects: 1) Using semantic features from the Vision Foundation Model (VFM), such as DINO, might limit generalization on unseen data due to reliance on prior knowledge. 2) The transient field requires significant memory to handle transient elements with perview Gaussians and struggles to define clear boundaries for occluders, solely relying on photometric errors. To address these problems, we propose ForestSplats, a novel approach that leverages the deformable transient field and a superpixel-aware mask to efficiently represent transient elements in the 2D scene across unconstrained image collections and effectively decompose static scenes from transient distractors without VFM. We designed the transient field to be deformable, capturing per-view transient elements. Furthermore, we introduce a superpixel-aware mask that clearly defines the boundaries of occluders by considering photometric errors and superpixels. Additionally, we propose uncertainty-aware densification to avoid generating Gaus-sians within the boundaries of occluders during densification. Through extensive experiments across several benchmark datasets, we demonstrate that ForestSplats outperforms existing methods without VFM and shows significant memory efficiency in representing transient elements.

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.

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.

In this paper, we propose methods to estimate the moving direction and size of a target vehicle based on point cloud data detected by high-resolution automotive radar sensor. Previous studies using automotive radar sensors have proposed methods to roughly estimate the moving direction of vehicles, such as left, straight, or right. This study proposes methods to estimate not only the specific moving direction but also the approximate size of the vehicle. First, we use the high-resolution frequency-modulated continuous wave radar to acquire point cloud data for vehicles moving at various angles. In the point cloud data, radar signals are strongly reflected from the side of the vehicle and detected as a line segment. The proposed moving direction and size estimation method is based on line segment extracted from the Hough transform (HT). To extract the line segment from the point cloud data, the Hough transform is used. Using the extracted line segment, methods for estimating the moving direction and size of the vehicle are proposed. And then, the quick hull algorithm is used to estimate the center point of the vehicle to match the position of the target in the coordinate system. Finally, the direction, width, and length of the vehicle estimated from the proposed methods show average errors of 1.94$^\circ$, 4.32%, and 6.32%, respectively. In addition, when compared to the conventional principal component analysis (PCA)-based method, our proposed method exhibits superior performance in terms of estimation accuracy. Moreover, we have validated the effectiveness of the proposed method even in scenarios with multiple vehicles and in noisy road environments.

In addition to its success in representation learning, contrastive learning is effective in image anomaly detection. Although contrastive learning depends significantly on data augmentation methods, time-series data augmentation for time-series anomaly detection is not investigated sufficiently. Additionally, although time-series data share a temporal context, the existing contrastive loss contrasts temporally related samples, in which deteriorated anomaly detection performance is observed on time-series data. Herein, we propose contrastive multivariate time-series anomaly detection (CTAD), a multivariate time-series anomaly detection framework that addresses these challenges by incorporating a one-class learning scheme into the contrastive loss based on meticulously designed time-series data augmentations. Specifically, we propose seven types of general time-series data augmentations to be applied variable- and point- wise, and provide guidance on data augmentation methods for contrastive time-series anomaly detection. The superiority of the one-class contrastive loss and the appropriate selection of time-series data augmentation allow CTAD to achieve outstanding performance in multiple datasets, even using a simple long short-term memory network. Furthermore, CTAD is robust to noise as it trains a noise-invariant network. This enables up to 47× faster and 20× more memory-efficient anomaly detection performance compared with existing methods while affording robustness, which are essential considerations in real-world applications.

Deep learning (DL) has been extensively adopted in many applications, including disease prediction. Most DL-based applications are executed on a cloud server because the DL models are too large and complicated to be executed on the client-side. De facto cloud-hosted inferences lead to privacy concerns regarding services that operate on personal medical data. Nevertheless, given the recent development of DL-based applications for health-diagnosis services, these applications have become a dominant means of healthcare support in our daily lives. To prevent the misuse of personal medical data, several techniques have been developed to preserve sensitive information, with a trade-off between privacy and utility. A simple method that offers privacy preservation and good prediction performance involves the deployment of a diagnostic method to the client side. However, doing so makes DL models more vulnerable to adversaries. To this end, we propose a deep private generative framework that guarantees user-data privacy while maintaining the original class information and protecting the models from reverse engineering. Experimentation with practical deep neural networks on benchmark disease datasets demonstrates that the proposed method decreases the mutual information between the original data and synthetic data by nearly 80% while preserving a prediction accuracy of nearly 95% of the original prediction accuracy.

Concept-based explanation aims to provide concise and human-understandable explanations of an image classifier. However, existing concept-based explanation methods typically require a significant amount of manually collected concept-annotated images. This is costly and runs the risk of human biases being involved in the explanation. In this paper, we propose Counterfactual explanation with text-driven concepts (CounTEX), where the concepts are defined only from text by leveraging a pretrained multimodal joint embedding space without additional concept-annotated datasets. A conceptual counterfactual explanation is generated with text-driven concepts. To utilize the text-driven concepts defined in the joint embedding space to interpret target classifier outcome, we present a novel projection scheme for mapping the two spaces with a simple yet effective implementation. We show that CounTEX generates faithful explanations that provide a semantic understanding of model decision rationale robust to human bias.

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.

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.

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.

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.

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.

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.