paper-with-me

Papers

Towards Lossless Dataset Distillation via Difficulty-Aligned Trajectory Matching

2023-10-09 · Ziyao Guo, Kai Wang, George Cazenavette, Hui Li, Kaipeng Zhang, Yang You

The ultimate goal of Dataset Distillation is to synthesize a small synthetic dataset such that a model trained on this synthetic set will perform equally well as a model trained on the full, real dataset. Until now, no method of Dataset Distillation has reached this completely lossless goal, in part due to the fact that previous methods only remain effective when the total number of synthetic samples is extremely small. Since only so much information can be contained in such a small number of samples, it seems that to achieve truly loss dataset distillation, we must develop a distillation method that remains effective as the size of the synthetic dataset grows. In this work, we present such an algorithm and elucidate why existing methods fail to generate larger, high-quality synthetic sets. Current state-of-the-art methods rely on trajectory-matching, or optimizing the synthetic data to induce similar long-term training dynamics as the real data. We empirically find that the training stage of the trajectories we choose to match (i.e., early or late) greatly affects the effectiveness of the distilled dataset. Specifically, early trajectories (where the teacher network learns easy patterns) work well for a low-cardinality synthetic set since there are fewer examples wherein to distribute the necessary information. Conversely, late trajectories (where the teacher network learns hard patterns) provide better signals for larger synthetic sets since there are now enough samples to represent the necessary complex patterns. Based on our findings, we propose to align the difficulty of the generated patterns with the size of the synthetic dataset. In doing so, we successfully scale trajectory matching-based methods to larger synthetic datasets, achieving lossless dataset distillation for the very first time. Code and distilled datasets are available at https://gzyaftermath.github.io/DATM.

📄 PDF Abstract BibTeX arXiv:2310.05773

Code (1)

nus-hpc-ai-lab/datm 공식 구현 pytorch

Tasks

Dataset Distillation

Methods 이 논문이 사용한 방법론

ALIGN In the ALIGN method, visual and language representations are jointly trained from noisy image alt-text data. The image and text encoders are learned via contrastive loss…

Similar Papers 제목 키워드 기반

Enhancing Dataset Distillation via Label Inconsistency Elimination and Learning Pattern Refinement

2024-10-17 · Chuhao Zhou, Chenxi Jiang, Yi Xie, Haozhi Cao 외

Dataset Distillation (DD) seeks to create a condensed dataset that, when used to train a model, enables the model to achieve performance similar to that of a model trained on the entire original dataset. It relieves the …

Dataset Distillation

Hyper-SD: Trajectory Segmented Consistency Model for Efficient Image Synthesis

2024-04-21 · Yuxi Ren, Xin Xia, Yanzuo Lu, Jiacheng Zhang 외

Recently, a series of diffusion-aware distillation algorithms have emerged to alleviate the computational overhead associated with the multi-step inference process of Diffusion Models (DMs). Current distillation techniqu…

Image Generation

SelMatch: Effectively Scaling Up Dataset Distillation via Selection-Based Initialization and Partial Updates by Trajectory Matching

2024-05-28 · Yongmin Lee, Hye Won Chung

Dataset distillation aims to synthesize a small number of images per class (IPC) from a large dataset to approximate full dataset training with minimal performance loss. While effective in very small IPC ranges, many dis…

Dataset Distillation

When Teacher Guidance Misleads: Reward-Aligned On-Policy Distillation

2026-08-28 · Siyuan Gan, Yuhan Li, Xiran Wang, Linjian Meng 외 arxiv

On-policy distillation (OPD) has recently emerged as a popular post-training paradigm for large language models (LLMs), providing an efficient way to transfer the knowledge and capabilities of teacher models into student…

Prioritize Alignment in Dataset Distillation

2024-08-06 · Zekai Li, Ziyao Guo, Wangbo Zhao, Tianle Zhang 외

Dataset Distillation aims to compress a large dataset into a significantly more compact, synthetic one without compromising the performance of the trained models. To achieve this, existing methods use the agent model to …

Dataset Distillation