arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2006.05929cs.CVcs.LG
收藏

使用梯度匹配的数据集压缩

Dataset Condensation with Gradient Matching

School of Informatics, The University of Edinburgh(爱丁堡大学信息学院)

Bo Zhao, Konda Reddy Mopuri, Hakan Bilen

摘要:

由于许多领域中最先进的机器学习方法依赖于更大的数据集,因此存储数据集并在其上训练模型变得更加昂贵。本文提出了一种用于数据高效学习的训练集合成技术,称为数据集浓缩,该技术学会将大型数据集浓缩为一小组信息丰富的合成样本,用于从头开始训练深度神经网络。我们将这个目标表述为在原始数据和合成数据上训练的深度神经网络权重梯度之间的梯度匹配问题。我们严格评估其性能在几个计算机视觉基准,并证明它显着优于国家的最先进的方法。最后,我们探索使用我们的方法在持续学习和神经结构搜索和报告有前途的收益时,有限的内存和计算。

英文摘要:

As the state-of-the-art machine learning methods in many fields rely on larger datasets, storing datasets and training models on them become significantly more expensive. This paper proposes a training set synthesis technique for data-efficient learning, called Dataset Condensation, that learns to condense large dataset into a small set of informative synthetic samples for training deep neural networks from scratch. We formulate this goal as a gradient matching problem between the gradients of deep neural network weights that are trained on the original and our synthetic data. We rigorously evaluate its performance in several computer vision benchmarks and demonstrate that it significantly outperforms the state-of-the-art methods. Finally we explore the use of our method in continual learning and neural architecture search and report promising gains when limited memory and computations are available.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑