A Proposal for Adaptive Synapse Allocation in Cortical Learning Algorithm with Dynamic Range

Abstract

The Cortical Learning Algorithm (CLA) is inspired by the human neocortex. It is robust to changes in data trends and well-suited for online time-series prediction. In this study, we propose a method to allocate synapses adaptively according to input data for CLA-DR, which removes predefined input range constraints by adding synapses. Experimental results demonstrate that the proposed method improves prediction accuracy by enabling more effective utilization of the model’s resources.