Task-Driven Convolutional Recurrent Models of the Visual System

TL;DR

This paper investigates how recurrence in convolutional neural networks can improve object recognition and better emulate neural activity in the primate visual system, introducing novel recurrent cell structures and feedback mechanisms.

Contribution

It introduces new recurrent cell designs with bypassing and gating, and demonstrates their effectiveness in enhancing CNN performance and neural plausibility.

Findings

Recurrent CNNs with novel cells outperform standard CNNs on ImageNet.

Task-optimized recurrent models better match primate neural dynamics.

Long-range feedback and local recurrence improve object recognition.

Abstract

Feed-forward convolutional neural networks (CNNs) are currently state-of-the-art for object classification tasks such as ImageNet. Further, they are quantitatively accurate models of temporally-averaged responses of neurons in the primate brain's visual system. However, biological visual systems have two ubiquitous architectural features not shared with typical CNNs: local recurrence within cortical areas, and long-range feedback from downstream areas to upstream areas. Here we explored the role of recurrence in improving classification performance. We found that standard forms of recurrence (vanilla RNNs and LSTMs) do not perform well within deep CNNs on the ImageNet task. In contrast, novel cells that incorporated two structural features, bypassing and gating, were able to boost task accuracy substantially. We extended these design principles in an automated search over thousands of model architectures, which identified novel local recurrent cells and long-range feedback connections useful for object recognition. Moreover, these task-optimized ConvRNNs matched the dynamics of neural activity in the primate visual system better than feedforward networks, suggesting a role for the brain's recurrent connections in performing difficult visual behaviors.