MMIM: An Interpretable Regularization Method for Neural Networks (Student Abstract)

Authors

  • Nan Xie Tianjin University
  • Yuexian Hou Tianjin University

Keywords:

Regularization Method, Mutual Information, Interpretability

Abstract

In deep learning models, most of network architectures are designed artificially and empirically. Although adding new structures such as convolution kernels in CNN is widely used, there are few methods to design new structures and mathematical tools to evaluate feature representation capabilities of new structures. Inspired by ensemble learning, we propose an interpretable regularization method named Minimize Mutual Information Method(MMIM), which minimize the generalization error by minimizing the mutual information of hidden neurons. The experimental results also verify the effectiveness of our proposed MMIM.

Downloads

Published

2021-05-18

How to Cite

Xie, N., & Hou, Y. (2021). MMIM: An Interpretable Regularization Method for Neural Networks (Student Abstract). Proceedings of the AAAI Conference on Artificial Intelligence, 35(18), 15933-15934. Retrieved from https://ojs.aaai.org/index.php/AAAI/article/view/17963

Issue

Section

AAAI Student Abstract and Poster Program