Deep Neural Networks with Trainable Activations and Controlled Lipschitz Constant

Abstract

We introduce a variational framework to learn the activation functions of deep neural networks. Our aim is to increase the capacity of the network while controlling an upper-bound of the actual Lipschitz constant of the input-output relation. To that end, we first establish a global bound for the Lipschitz constant of neural networks. Based on the obtained bound, we then formulate a variational problem for learning activation functions. Our variational problem is infinite-dimensional and is not computationally tractable. However, we prove that there always exists a solution that has continuous and piecewise-linear (linear-spline) activations. This reduces the original problem to a finite-dimensional minimization where an l1 penalty on the parameters of the activations favors the learning of sparse nonlinearities. We numerically compare our scheme with standard ReLU network and its variations, PReLU and LeakyReLU and we empirically demonstrate the practical aspects of our framework.

References in corpus (8)

• Deep Convolutional Neural Network for Inverse Problems in Imaging
• On the Number of Linear Regions of Deep Neural Networks
• On instabilities of deep learning in image reconstruction - Does AI come at a cost?
• CNN-Based Projected Gradient Descent for Consistent Image Reconstruction
• Learning Activation Functions to Improve Deep Neural Networks
• Compressed Sensing using Generative Models
• Understanding Deep Neural Networks with Rectified Linear Units
• Splines are Universal Solutions of Linear Inverse Problems with Generalized-TV regularization

Cited by in corpus (4)

• What Kinds of Functions do Deep Neural Networks Learn? Insights from Variational Spline Theory
• CLIP: Cheap Lipschitz Training of Neural Networks
• TV-based Reconstruction of Periodic Functions
• Learning Lipschitz Feedback Policies from Expert Demonstrations: Closed-Loop Guarantees, Generalization and Robustness