Dataset condensation with contrastive signals
WebSep 12, 2024 · In this work, we analyse the contrastive fine-tuning of pre-trained language models on two fine-grained text classification tasks, emotion classification and sentiment analysis. We adaptively embed class relationships into a contrastive objective function to help differently weigh the positives and negatives, and in particular, weighting ... WebFigure 1: Dataset Condensation (left) aims to generate a small set of synthetic images that can match the performance of a network trained on a large image dataset. Our method (right) realizes this goal by learning a synthetic set such that a deep network trained on it and the large set produces similar gradients w.r.t. its weights.
Dataset condensation with contrastive signals
Did you know?
WebTo 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 ... WebCurrently, he works as the head of NAVER AI Lab in NAVER Cloud. He has contributed to the AI research community as Datasets and Benchmarks Co-chair for NeurIPS and Social Co-chair for ICML 2024 and NeurIPS 2024. Also, he has joined a senior technical program committee member, such as, Area chair for NeurIPS 2024 and 2024, Area chair for ICML ...
WebJun 4, 2024 · Dataset Condensation with Contrastive Signals Recent studies have demonstrated that gradient matching-based dataset sy... 0 Saehyung Lee, et al. ∙. share ... WebTitle: Dataset Condensation with Contrastive Signals Authors: Saehyung Lee, Sanghyuk Chun, Sangwon Jung, Sangdoo Yun, Sungroh Yoon Abstract summary: gradient …
WebDataset Condensation with Contrastive Signals (Saehyung Lee et al., ICML 2024) 📖 Delving into Effective Gradient Matching for Dataset Condensation (Zixuan Jiang et al., 2024) 📖 Scaling Up Dataset Distillation to ImageNet-1K with Constant Memory (Justin Cui et … Web[24]Saehyung Lee, Sanghyuk Chun, Sangwon Jung, Sangdoo Yun, Sungroh Yoon, \Dataset Condensation with Contrastive Signals", International Conference on Machine Learning (ICML), 2024. ... IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2024 [7]Sangdoo Yun, Dongyoon Han, Seong Joon Oh, …
WebFeb 6, 2024 · To address this problem, we propose Dataset Condensation with Contrastive signals (DCC) by modifying the loss function to enable the DC methods to …
shanghai 2 o\u0027clockWebHa, Hyun Oh Song, "Dataset Condensation via Efficient Synthetic-Data Parameterization", Interna-tional Conference on Machine Learning (ICML 2024), 2024. … shanghai 2 tischWebRecent studies on dataset condensation attempt to reduce the dependence on such massive data by synthesizing a compact training dataset. However, the existing … shanghai 30\\u0027s oxfordWebDataset Condensation With Contrastive Signals relevant information (e.g., logo, police sign, trailers) while suppressing task-irrelevant information (e.g., wheels, head … shanghai 24 hour visaWebConclusion •We show that DC primarily focuses on the class-wise gradient while overlooking contrastive signals. •To address this issue, we propose the Dataset Condensation with Contrastive signals (DCC) method. •In our experiments, we demonstrate that the proposed DCC outperforms DC in fine-grained classification tasks and general benchmark datasets shanghai 2 taipei queenstownWebDataset Condensation with Contrastive Signals. Contribute to Daankrol/DCC development by creating an account on GitHub. shanghai 30\u0027s oxfordWebJul 24, 2024 · Online Continual Learning with Contrastive Vision Transformer. Online continual learning (online CL) studies the problem of learning sequential tasks from an … shanghai 2 o\\u0027clock