We discuss relations between Residual Networks (ResNet), Recurrent Neural Networks (RNNs) and the primate visual cortex. We begin with the observation that a special type of shallow RNN is exactly equivalent to a very deep ResNet with weight sharing among the layers. A direct implementation of such a RNN, although having orders of magnitude fewer parameters, leads to a performance similar to the corresponding ResNet. We propose 1) a generalization of both RNN and ResNet architectures and 2) the conjecture that a class of moderately deep RNNs is a biologically-plausible model of the ventral stream in visual cortex. We demonstrate the effectiveness of the architectures by testing them on the CIFAR-10 and ImageNet dataset.
@article{arxiv.1604.03640,
title = {Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortex},
author = {Qianli Liao and Tomaso Poggio},
journal= {arXiv preprint arXiv:1604.03640},
year = {2021}
}
Comments
This version was written in Sept. 2016. For April 2016 version see v1 below