arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.03314cs.LGstat.ML
收藏

用最优传输度量泛化

Measuring Generalization with Optimal Transport

MIT CSAIL(麻省理工学院计算机科学与人工智能实验室);IBM Research AI(IBM人工智能研究院);MIT-IBM Watson AI Lab(麻省理工学院-IBM沃森人工智能实验室)

Ching-Yao Chuang, Youssef Mroueh, Kristjan Greenewald, Antonio Torralba, Stefanie Jegelka

摘要:

理解深度神经网络的泛化是深度学习中最重要的任务之一。虽然已经取得了很大的进展,理论误差界仍然经常表现出与经验观察。在这项工作中,我们开发了基于边际的泛化边界,其中边际被归一化为从训练分布中采样的独立随机子集之间的最佳运输成本。特别是,最优传输成本可以被解释为方差的泛化,它捕获了学习特征空间的结构属性。在给定训练数据和网络参数的情况下,我们的界在大规模数据集上鲁棒地预测泛化误差。理论上,我们证明了特征的集中和分离在泛化中起着至关重要的作用,支持文献中的实证结果。代码可在\url{https://github.com/chingyaoc/kV-Margin}上获得。

英文摘要:

Understanding the generalization of deep neural networks is one of the most important tasks in deep learning. Although much progress has been made, theoretical error bounds still often behave disparately from empirical observations. In this work, we develop margin-based generalization bounds, where the margins are normalized with optimal transport costs between independent random subsets sampled from the training distribution. In particular, the optimal transport cost can be interpreted as a generalization of variance which captures the structural properties of the learned feature space. Our bounds robustly predict the generalization error, given training data and network parameters, on large scale datasets. Theoretically, we demonstrate that the concentration and separation of features play crucial roles in generalization, supporting empirical results in the literature. The code is available at \url{https://github.com/chingyaoc/kV-Margin}.

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。

↑