The Uncanny Similarity of Recurrence and Depth
This work addresses the efficiency and capability of recurrent networks for researchers in deep learning, though it is incremental as it builds on existing knowledge of network depth and parameter reuse.
The study tackled the problem of understanding hierarchical feature extraction in recurrent networks compared to feed-forward networks, showing that recurrent models achieve similar hierarchical behaviors and performance benefits with depth while using fewer parameters.
It is widely believed that deep neural networks contain layer specialization, wherein neural networks extract hierarchical features representing edges and patterns in shallow layers and complete objects in deeper layers. Unlike common feed-forward models that have distinct filters at each layer, recurrent networks reuse the same parameters at various depths. In this work, we observe that recurrent models exhibit the same hierarchical behaviors and the same performance benefits with depth as feed-forward networks despite reusing the same filters at every recurrence. By training models of various feed-forward and recurrent architectures on several datasets for image classification as well as maze solving, we show that recurrent networks have the ability to closely emulate the behavior of non-recurrent deep models, often doing so with far fewer parameters.