Informed Learning by Wide Neural Networks: Convergence, Generalization and Sampling Complexity

TL;DR

This paper provides a theoretical analysis of how domain knowledge enhances deep neural network training by regularizing supervision and supplementing data, offering insights into convergence, generalization, and sampling complexity.

Contribution

It introduces a theoretical framework for informed deep learning, quantifies the benefits of domain knowledge, and proposes a new training objective to optimize knowledge integration.

Findings

Domain knowledge regularizes supervision and supplements data.

Trade-off exists between label and knowledge imperfectness.

Proposed objective improves population risk bounds.

Abstract

By integrating domain knowledge with labeled samples, informed machine learning has been emerging to improve the learning performance for a wide range of applications. Nonetheless, rigorous understanding of the role of injected domain knowledge has been under-explored. In this paper, we consider an informed deep neural network (DNN) with over-parameterization and domain knowledge integrated into its training objective function, and study how and why domain knowledge benefits the performance. Concretely, we quantitatively demonstrate the two benefits of domain knowledge in informed learning - regularizing the label-based supervision and supplementing the labeled samples - and reveal the trade-off between label and knowledge imperfectness in the bound of the population risk. Based on the theoretical analysis, we propose a generalized informed training objective to better exploit the benefits of knowledge and balance the label and knowledge imperfectness, which is validated by the population risk bound. Our analysis on sampling complexity sheds lights on how to choose the hyper-parameters for informed learning, and further justifies the advantages of knowledge informed learning.