Robust Learning of Recurrent Neural Networks in Presence of Exogenous Noise

TL;DR

This paper introduces a control-theoretic approach to analyze and improve the robustness of recurrent neural networks against input noise, demonstrating significant robustness enhancements through a new learning method.

Contribution

It proposes a tractable robustness analysis framework for RNNs using control theory and develops a learning method to enhance their robustness against Gaussian noise.

Findings

Robustness measure can be efficiently estimated via linearization.

The proposed method significantly improves RNN robustness in simulations.

The approach effectively quantifies the impact of input noise on RNN performance.

Abstract

Recurrent Neural networks (RNN) have shown promising potential for learning dynamics of sequential data. However, artificial neural networks are known to exhibit poor robustness in presence of input noise, where the sequential architecture of RNNs exacerbates the problem. In this paper, we will use ideas from control and estimation theories to propose a tractable robustness analysis for RNN models that are subject to input noise. The variance of the output of the noisy system is adopted as a robustness measure to quantify the impact of noise on learning. It is shown that the robustness measure can be estimated efficiently using linearization techniques. Using these results, we proposed a learning method to enhance robustness of a RNN with respect to exogenous Gaussian noise with known statistics. Our extensive simulations on benchmark problems reveal that our proposed methodology significantly improves robustness of recurrent neural networks.