Training Normalizing Flows with the Information Bottleneck for Competitive Generative Classification

Abstract

The Information Bottleneck (IB) objective uses information theory to formulate a task-performance versus robustness trade-off. It has been successfully applied in the standard discriminative classification setting. We pose the question whether the IB can also be used to train generative likelihood models such as normalizing flows. Since normalizing flows use invertible network architectures (INNs), they are information-preserving by construction. This seems contradictory to the idea of a bottleneck. In this work, firstly, we develop the theory and methodology of IB-INNs, a class of conditional normalizing flows where INNs are trained using the IB objective: Introducing a small amount of {\em controlled} information loss allows for an asymptotically exact formulation of the IB, while keeping the INN's generative capabilities intact. Secondly, we investigate the properties of these models experimentally, specifically used as generative classifiers. This model class offers advantages such as improved uncertainty quantification and out-of-distribution detection, but traditional generative classifier solutions suffer considerably in classification accuracy. We find the trade-off parameter in the IB controls a mix of generative capabilities and accuracy close to standard classifiers. Empirically, our uncertainty estimates in this mixed regime compare favourably to conventional generative and discriminative classifiers.

References in corpus (13)

• The information bottleneck method
• On Calibration of Modern Neural Networks
• Semi-Supervised Learning with Deep Generative Models
• Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift
• Opening the Black Box of Deep Neural Networks via Information
• A Neural Representation of Sketch Drawings
• MINE: Mutual Information Neural Estimation
• Towards the first adversarially robust neural network model on MNIST
• Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One
• Coupling-based Invertible Neural Networks Are Universal Diffeomorphism Approximators
• Sum-of-Squares Polynomial Flow
• Hybrid Models with Deep and Invertible Features
• A Theory of Usable Information Under Computational Constraints