Input Convex Gradient Networks

Abstract

The gradients of convex functions are expressive models of non-trivial vector fields. For example, Brenier's theorem yields that the optimal transport map between any two measures on Euclidean space under the squared distance is realized as a convex gradient, which is a key insight used in recent generative flow models. In this paper, we study how to model convex gradients by integrating a Jacobian-vector product parameterized by a neural network, which we call the Input Convex Gradient Network (ICGN). We theoretically study ICGNs and compare them to taking the gradient of an Input-Convex Neural Network (ICNN), empirically demonstrating that a single layer ICGN can fit a toy example better than a single layer ICNN. Lastly, we explore extensions to deeper networks and connections to constructions from Riemannian geometry.

arXiv commentAccepted to NeurIPS 2021 Optimal Transport and Machine Learning Workshop https://otml2021.github.io

References in corpus (11)

• Deep Equilibrium Models
• Input Convex Neural Networks
• Sobolev Training for Neural Networks
• Wasserstein-2 Generative Networks
• Scalable Computations of Wasserstein Barycenter via Input Convex Neural Networks
• Primal-Dual Wasserstein GAN
• Large-Scale Wasserstein Gradient Flows
• On approximating $\nabla f$ with neural networks
• Optimizing Functionals on the Space of Probabilities with Input Convex Neural Networks
• Neural Spatio-Temporal Point Processes
• Gradients are Not All You Need