Implicitly Defined Layers in Neural Networks

Abstract

In conventional formulations of multilayer feedforward neural networks, the individual layers are customarily defined by explicit functions. In this paper we demonstrate that defining individual layers in a neural network \emph{implicitly} provide much richer representations over the standard explicit one, consequently enabling a vastly broader class of end-to-end trainable architectures. We present a general framework of implicitly defined layers, where much of the theoretical analysis of such layers can be addressed through the implicit function theorem. We also show how implicitly defined layers can be seamlessly incorporated into existing machine learning libraries. In particular with respect to current automatic differentiation techniques for use in backpropagation based training. Finally, we demonstrate the versatility and relevance of our proposed approach on a number of diverse example problems with promising results.

References in corpus (6)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Task-Driven Dictionary Learning
• Composing graphical models with neural networks for structured representations and fast inference
• Input Convex Neural Networks
• On Differentiating Parameterized Argmin and Argmax Problems with Application to Bi-level Optimization
• Matryoshka Networks: Predicting 3D Geometry via Nested Shape Layers

Cited by in corpus (2)

• Differentiable Implicit Layers
• Implicit Riemannian Concave Potential Maps