检索结果-内蒙古大学图书馆

Textual Training for the Hassle-Free Removal of Unwanted Visual Data: Case Studies on OOD and Hateful Image Detection

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lee, Saehyung Mok, Jisoo Park, Sangha Shin, Yongho Jung, Dahuin Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of Qualcomm Korea YH Seoul Korea Republic of School of Computer Science and Engineering Soongsil University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of

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 synthetic 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 out-of-distribution detection and hateful image detection. Our codes are available at https://***/Saehyung-Lee/HFTT © 2024, CC BY.

关键词： Data assimilation

DAFA: DISTANCE-AWARE FaiR ADVERSARIAL TRaiNING

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lee, Hyungyu Lee, Saehyung Jang, Hyemi Park, Junsung Bae, Ho Yoon, Sungroh Electrical and Computer Engineering Seoul National University Korea Republic of Department of Cyber Security Ewha Womans University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of

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 loss weights and adversarial margins 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. Copyright © 2024, The Authors. All rights reserved.

关键词： Economic and social effects

ProPILE: Probing Privacy Leakage in Large Language Models

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Kim, Siwon Yun, Sangdoo Lee, Hwaran Gubri, Martin Yoon, Sungroh Oh, Seong Joon Department of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of NAVER AI Lab Korea Republic of University of Luxembourg Luxembourg Parameter Lab Germany Tübingen AI Center University of Tübingen Germany

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. © 2023, CC BY-NC-SA.

关键词： Piles

Gradient Alignment with Prototype Feature for Fully Test-time Adaptation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Shin, Juhyeon Lee, Jonghyun Lee, Saehyung Park, Minjun Lee, Dongjun Hwang, Uiwon Yoon, Sungroh Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of Department of Electrical and Computer Engineering Seoul National University Korea Republic of Division of Digital Healthcare Yonsei University Korea Republic of

In context of Test-time Adaptation(TTA), we propose a regularizer, dubbed Gradient Alignment with Prototype feature (GAP), which alleviates the inappropriate guidance from entropy minimization loss from misclassified pseudo label. We developed a gradient alignment loss to precisely manage the adaptation process, ensuring that changes made for some data don’t negatively impact the model’s performance on other data. We introduce a prototype feature of a class as a proxy measure of the negative impact. To make GAP regularizer feasible under the TTA constraints, where model can only access test data without labels, we tailored its formula in two ways: approximating prototype features with weight vectors of the classifier, calculating gradient without back-propagation. We demonstrate GAP significantly improves TTA methods across various datasets, which proves its versatility and effectiveness. Copyright © 2024, The Authors. All rights reserved.

关键词： Alignment

Introducing Spectral Attention for Long-Range Dependency in Time Series Forecasting

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Kang, Bong Gyun Lee, Dongjun Kim, HyunGi Yoon, Sungroh Chung, DoHyun Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of Department of Electrical and Computer Engineering Seoul National University Korea Republic of Department of Future Automotive Mobility Seoul National University Korea Republic of

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. © 2024, CC BY-NC-ND.

关键词： Time series

SF(DA)2: SOURCE-FREE DOMaiN ADAPTATION THROUGH THE LENS OF DATA AUGMENTATION

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hwang, Uiwon Lee, Jonghyun Shin, Juhyeon Yoon, Sungroh Division of Digital Healthcare Yonsei University Korea Republic of Department of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of

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. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantics

On mitigating stability-plasticity dilemma in CLIP-guided image morphing via geodesic distillation loss

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Oh, Yeongtak Lee, Saehyung Hwang, Uiwon Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of Division of Digital Healthcare Yonsei University Korea Republic of

Large-scale language-vision pre-training models, such as CLIP, have achieved remarkable text-guided image morphing results by leveraging several unconditional generative models. However, existing CLIP-guided image morphing methods encounter difficulties when morphing photorealistic images. Specifically, existing guidance fails to provide detailed explanations of the morphing regions within the image, leading to misguidance. In this paper, we observed that such misguidance could be effectively mitigated by simply using a proper regularization loss. Our approach comprises two key components: 1) a geodesic cosine similarity loss that minimizes inter-modality features (i.e., image and text) on a projected subspace of CLIP space, and 2) a latent regularization loss that minimizes intra-modality features (i.e., image and image) on the image manifold. By replacing the naïve directional CLIP loss in a drop-in replacement manner, our method achieves superior morphing results on both images and videos for various benchmarks, including CLIP-inversion. © 2024, CC BY.

关键词： Distillation

Towards High Generalization Performance on Electrocardiogram Classification

学校读者我要写书评

暂无评论

Towards High Generalization Performance on Electrocardiogram...

2021 Computing in Cardiology, CinC 2021

作者： Han, Hyeongrok Park, Seongjae Min, Seonwoo Choi, Hyun-Soo Kim, Eunji Kim, Hyunki Park, Sangha Kim, Jinkook Park, Junsang An, Junho Lee, Kwanglo Jeong, Wonsun Chon, Sangil Ha, Kwonwoo Han, Myungkyu Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Seoul Korea Republic of HUINNO Co. Ltd Seoul Korea Republic of LG AI Research Seoul Korea Republic of Department of Computer Science and Engineering Kangwon National University Chuncheon Korea Republic of Department of Biological Sciences Interdisciplinary Program in Bioinformatics Interdisciplinary Program in Artificial Intelligence ASRI INMC Institute of Engineering Research Seoul National University Seoul Korea Republic of

ISBN: (纸本)9781665479165

Recently, many electrocardiogram (ECG) classification algorithms using deep learning have been proposed. The characteristics of ECG vary from dataset to dataset for various reasons (i.e., hospital, race, etc.). Therefore, it is important for a model to have high generalization performance consistently over all datasets. In this paper, as part of the PhysioNet / Computing in Cardiology Challenge 2021, we present a model developed to classify cardiac abnormalities from 12 lead and reduced-lead ECGs. In particular, to upgrade our previous model for improving generalization performance, we newly adopt constant-weighted cross-entropy loss, additional features, Mixup augmentation, and squeeze/excitation block, OneCycle learning rate scheduler, which are selected via evaluation of generalization performance using leave-one-dataset-out cross-validation setting. With the present model, our DSaiLSNU team has received challenge scores of 0.55, 0.58, 0.58, 0.57 and 0.57 (ranked 2nd, 1st, 1st, 2nd, 2nd out of 39 teams) for the 12-lead, 6-lead, 4-lead, 3-lead, and 2-lead versions of the hidden test set, respectively. The present model achieves higher generalization performance over all versions of the hidden test set than the model submitted last year. © 2021 Creative Commons.

关键词： Deep learning

Efficient Diffusion-Driven Corruption Editor for Test-Time Adaptation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Oh, Yeongtak Lee, Jonghyun Choi, Jooyoung Jung, Dahuin Hwang, Uiwon Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of School of Computer Science and Engineering Soongsil University Korea Republic of Division of Digital Healthcare Yonsei University Korea Republic of

Test-time adaptation (TTA) addresses the unforeseen distribution shifts occurring during test time. In TTA, performance, memory consumption, and time consumption are crucial considerations. A recent diffusion-based TTA approach for restoring corrupted images involves image-level updates. However, using pixel space diffusion significantly increases resource requirements compared to conventional model updating TTA approaches, revealing limitations as a TTA method. To address this, we propose a novel TTA method that leverages an image editing model based on a latent diffusion model (LDM) and fine-tunes it using our newly introduced corruption modeling scheme. This scheme enhances the robustness of the diffusion model against distribution shifts by creating (clean, corrupted) image pairs and fine-tuning the model to edit corrupted images into clean ones. Moreover, we introduce a distilled variant to accelerate the model for corruption editing using only 4 network function evaluations (NFEs). We extensively validated our method across various architectures and datasets including image and video domains. Our model achieves the best performance with a 100 times faster runtime than that of a diffusion-based baseline. Furthermore, it is three times faster than the previous model updating TTA method that utilizes data augmentation, making an image-level updating approach more feasible. ∗ © 2024, CC BY.

关键词：