Sobolev Training for Neural Networks

Abstract

At the heart of deep learning we aim to use neural networks as function approximators - training them to produce outputs from inputs in emulation of a ground truth function or data creation process. In many cases we only have access to input-output pairs from the ground truth, however it is becoming more common to have access to derivatives of the target output with respect to the input - for example when the ground truth function is itself a neural network such as in network compression or distillation. Generally these target derivatives are not computed, or are ignored. This paper introduces Sobolev Training for neural networks, which is a method for incorporating these target derivatives in addition the to target values while training. By optimising neural networks to not only approximate the function's outputs but also the function's derivatives we encode additional information about the target function within the parameters of the neural network. Thereby we can improve the quality of our predictors, as well as the data-efficiency and generalization capabilities of our learned function approximation. We provide theoretical justifications for such an approach as well as examples of empirical evidence on three distinct domains: regression on classical optimisation datasets, distilling policies of an agent playing Atari, and on large-scale applications of synthetic gradients. In all three domains the use of Sobolev Training, employing target derivatives in addition to target values, results in models with higher accuracy and stronger generalisation.

References in corpus (8)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• Playing Atari with Deep Reinforcement Learning
• Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
• Deep Model Compression: Distilling Knowledge from Noisy Teachers
• Decoupled Neural Interfaces using Synthetic Gradients
• Exploiting gradients and Hessians in Bayesian optimization and Bayesian quadrature

Cited by in corpus (8)

• Geometric deep learning for computational mechanics Part I: Anisotropic Hyperelasticity
• Adaptive weighting of Bayesian physics informed neural networks for multitask and multiscale forward and inverse problems
• What machine learning can do for computational solid mechanics
• Grappa -- A Machine Learned Molecular Mechanics Force Field
• Label-guided Attention Distillation for Lane Segmentation
• Physics-augmented neural networks for constitutive modeling of hyperelastic geometrically exact beams
• Nonclosedness of Sets of Neural Networks in Sobolev Spaces
• Input Convex Gradient Networks