Recombinant dynamical systems

TL;DR

This paper introduces a connectionist model that mimics experience-based problem solving by recombining prior solutions, applied to efficient sequence generation, demonstrating promising empirical results.

Contribution

It presents a novel connectionist approach for recombining solutions to solve new problems without relying on gradient descent.

Findings

Empirical tests show the model's utility in sequence generation.

The model effectively recombines prior solutions for new problems.

Results indicate potential for solving problems lacking gradient descent methods.

Abstract

We describe a connectionist model that attempts to capture a notion of experience-based problem solving or task learning, whereby solutions to newly encountered problems are composed from remembered solutions to prior problems. We apply this model to the computational problem of \emph{efficient sequence generation}, a problem for which there is no obvious gradient descent procedure, and for which not all posable problem instances are solvable. Empirical tests show promising evidence of utility.