Neural Networks with Recurrent Generative Feedback

TL;DR

This paper introduces a novel CNN architecture with generative recurrent feedback inspired by the Bayesian brain hypothesis, significantly enhancing adversarial robustness by enforcing self-consistency through Bayesian MAP inference.

Contribution

The paper proposes CNN-F, a new neural network framework that incorporates generative feedback and Bayesian inference to improve robustness against adversarial attacks.

Findings

CNN-F outperforms standard CNNs in adversarial robustness.

Generative feedback enforces self-consistency, leading to more stable predictions.

Experimental results on benchmarks demonstrate significant robustness improvements.

Abstract

Neural networks are vulnerable to input perturbations such as additive noise and adversarial attacks. In contrast, human perception is much more robust to such perturbations. The Bayesian brain hypothesis states that human brains use an internal generative model to update the posterior beliefs of the sensory input. This mechanism can be interpreted as a form of self-consistency between the maximum a posteriori (MAP) estimation of an internal generative model and the external environment. Inspired by such hypothesis, we enforce self-consistency in neural networks by incorporating generative recurrent feedback. We instantiate this design on convolutional neural networks (CNNs). The proposed framework, termed Convolutional Neural Networks with Feedback (CNN-F), introduces a generative feedback with latent variables to existing CNN architectures, where consistent predictions are made through alternating MAP inference under a Bayesian framework. In the experiments, CNN-F shows considerably improved adversarial robustness over conventional feedforward CNNs on standard benchmarks.