Recurrent Online Clustering as a Spatio-Temporal Feature Extractor in DeSTIN

TL;DR

This paper enhances the DeSTIN architecture by replacing transition tables with a feedback-based clustering mechanism, leading to state-of-the-art MNIST classification results.

Contribution

It introduces a scalable, feedback-driven clustering method for DeSTIN that improves temporal feature extraction without explicit transition tables.

Findings

Achieved state-of-the-art MNIST classification accuracy

Simplified the DeSTIN architecture with feedback-based clustering

Demonstrated improved scalability and performance

Abstract

This paper presents a basic enhancement to the DeSTIN deep learning architecture by replacing the explicitly calculated transition tables that are used to capture temporal features with a simpler, more scalable mechanism. This mechanism uses feedback of state information to cluster over a space comprised of both the spatial input and the current state. The resulting architecture achieves state-of-the-art results on the MNIST classification benchmark.