Learning Optimal Representations with the Decodable Information Bottleneck
arXiv:2009.12789
Abstract
We address the question of characterizing and finding optimal representations for supervised learning. Traditionally, this question has been tackled using the Information Bottleneck, which compresses the inputs while retaining information about the targets, in a decoder-agnostic fashion. In machine learning, however, our goal is not compression but rather generalization, which is intimately linked to the predictive family or decoder of interest (e.g. linear classifier). We propose the Decodable Information Bottleneck (DIB) that considers information retention and compression from the perspective of the desired predictive family. As a result, DIB gives rise to representations that are optimal in terms of expected test performance and can be estimated with guarantees. Empirically, we show that the framework can be used to enforce a small generalization gap on downstream classifiers and to predict the generalization ability of neural networks.
Accepted at NeurIPS 2020
References in corpus (14)
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Learning deep representations by mutual information estimation and maximization
- Gradient-based Hyperparameter Optimization through Reversible Learning
- Regularizing Neural Networks by Penalizing Confident Output Distributions
- On Mutual Information Maximization for Representation Learning
- Path-SGD: Path-Normalized Optimization in Deep Neural Networks
- Predicting the Generalization Gap in Deep Networks with Margin Distributions
- Generalized Inner Loop Meta-Learning
- Norm-Based Capacity Control in Neural Networks
- Data-dependent Sample Complexity of Deep Neural Networks via Lipschitz Augmentation
- A Theory of Usable Information Under Computational Constraints
- Understanding Generalization through Visualizations
- Minimal Achievable Sufficient Statistic Learning
- An Overview on Data Representation Learning: From Traditional Feature Learning to Recent Deep Learning