Measuring Generalization with Optimal Transport
arXiv:2106.03314
Abstract
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}.
NeurIPS 2021
References in corpus (10)
- Scikit-learn: Machine Learning in Python
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Towards Understanding the Role of Over-Parametrization in Generalization of Neural Networks
- Uniform convergence may be unable to explain generalization in deep learning
- Generalization in Deep Networks: The Role of Distance from Initialization
- NeurIPS 2020 Competition: Predicting Generalization in Deep Learning
- Representation Based Complexity Measures for Predicting Generalization in Deep Learning
- Wasserstein Distributionally Robust Optimization and Variation Regularization
- Ranking Deep Learning Generalization using Label Variation in Latent Geometry Graphs
- Gi and Pal Scores: Deep Neural Network Generalization Statistics