Persistent learning signals and working memory without continuous attractors
This work addresses the challenge of designing artificial learning systems and understanding biological neural dynamics for temporal dependence, offering a novel approach that could impact both fields.
The paper tackles the problem of learning temporal relationships and working memory by showing that quasi-periodic attractors, unlike continuous attractors, can support learning without fine-tuning issues, and demonstrates this with a new initialization scheme for recurrent neural networks that outperforms standard methods.
Neural dynamical systems with stable attractor structures, such as point attractors and continuous attractors, are hypothesized to underlie meaningful temporal behavior that requires working memory. However, working memory may not support useful learning signals necessary to adapt to changes in the temporal structure of the environment. We show that in addition to the continuous attractors that are widely implicated, periodic and quasi-periodic attractors can also support learning arbitrarily long temporal relationships. Unlike the continuous attractors that suffer from the fine-tuning problem, the less explored quasi-periodic attractors are uniquely qualified for learning to produce temporally structured behavior. Our theory has broad implications for the design of artificial learning systems and makes predictions about observable signatures of biological neural dynamics that can support temporal dependence learning and working memory. Based on our theory, we developed a new initialization scheme for artificial recurrent neural networks that outperforms standard methods for tasks that require learning temporal dynamics. Moreover, we propose a robust recurrent memory mechanism for integrating and maintaining head direction without a ring attractor.