Approximating Stacked and Bidirectional Recurrent Architectures with the Delayed Recurrent Neural Network

TL;DR

This paper introduces the delayed-RNN, a single-layer RNN that can emulate the capabilities of stacked and bidirectional RNNs, offering similar or improved performance with potentially faster runtimes.

Contribution

The work proves that a weight-constrained delayed-RNN is equivalent to stacked RNNs and demonstrates its ability to mimic bidirectional networks, providing a simpler alternative.

Findings

Delayed-RNN can replicate bidirectional RNNs in acausal tasks.

Delayed-RNN outperforms bidirectional RNNs on some tasks.

Delayed-RNN achieves similar performance to bidirectional RNNs in NLP applications.

Abstract

Recent work has shown that topological enhancements to recurrent neural networks (RNNs) can increase their expressiveness and representational capacity. Two popular enhancements are stacked RNNs, which increases the capacity for learning non-linear functions, and bidirectional processing, which exploits acausal information in a sequence. In this work, we explore the delayed-RNN, which is a single-layer RNN that has a delay between the input and output. We prove that a weight-constrained version of the delayed-RNN is equivalent to a stacked-RNN. We also show that the delay gives rise to partial acausality, much like bidirectional networks. Synthetic experiments confirm that the delayed-RNN can mimic bidirectional networks, solving some acausal tasks similarly, and outperforming them in others. Moreover, we show similar performance to bidirectional networks in a real-world natural language processing task. These results suggest that delayed-RNNs can approximate topologies including stacked RNNs, bidirectional RNNs, and stacked bidirectional RNNs - but with equivalent or faster runtimes for the delayed-RNNs.