Continual Learning of Recurrent Neural Networks by Locally Aligning Distributed Representations

TL;DR

This paper introduces P-TNCN, a biologically inspired recurrent neural network training method that avoids back-propagation through time, enabling efficient, parallelizable learning, zero-shot adaptation, and improved sequence modeling performance.

Contribution

The paper proposes the Local Representation Alignment algorithm for training recurrent networks without unrolling or derivatives, outperforming traditional methods on sequence benchmarks.

Findings

Outperforms back-propagation through time on sequence tasks

Enables zero-shot adaptation with fixed weights

Facilitates online continual learning in recurrent networks

Abstract

Temporal models based on recurrent neural networks have proven to be quite powerful in a wide variety of applications. However, training these models often relies on back-propagation through time, which entails unfolding the network over many time steps, making the process of conducting credit assignment considerably more challenging. Furthermore, the nature of back-propagation itself does not permit the use of non-differentiable activation functions and is inherently sequential, making parallelization of the underlying training process difficult. Here, we propose the Parallel Temporal Neural Coding Network (P-TNCN), a biologically inspired model trained by the learning algorithm we call Local Representation Alignment. It aims to resolve the difficulties and problems that plague recurrent networks trained by back-propagation through time. The architecture requires neither unrolling in time nor the derivatives of its internal activation functions. We compare our model and learning procedure to other back-propagation through time alternatives (which also tend to be computationally expensive), including real-time recurrent learning, echo state networks, and unbiased online recurrent optimization. We show that it outperforms these on sequence modeling benchmarks such as Bouncing MNIST, a new benchmark we denote as Bouncing NotMNIST, and Penn Treebank. Notably, our approach can in some instances outperform full back-propagation through time as well as variants such as sparse attentive back-tracking. Significantly, the hidden unit correction phase of P-TNCN allows it to adapt to new datasets even if its synaptic weights are held fixed (zero-shot adaptation) and facilitates retention of prior generative knowledge when faced with a task sequence. We present results that show the P-TNCN's ability to conduct zero-shot adaptation and online continual sequence modeling.