Episodic Memory Theory for the Mechanistic Interpretation of Recurrent Neural Networks

no code implementations • 3 Oct 2023 • Arjun Karuvally, Peter DelMastro, Hava T. Siegelmann

Utilizing the EMT, we formulate a mathematically rigorous circuit that facilitates variable binding in these tasks.

On the Dynamics of Learning Time-Aware Behavior with Recurrent Neural Networks

no code implementations • 12 Jun 2023 • Peter DelMastro, Rushiv Arora, Edward Rietman, Hava T. Siegelmann

In this way, we demonstrate how dynamical systems theory can provide insights into not only the learned representations of these models, but also the dynamics of the learning process itself.