Self-adaptive node-based PCA encodings

TL;DR

This paper introduces Simple Hebbian PCA, a distributed algorithm for PCA that simplifies network structures by eliminating intralayer weights, reducing training complexity.

Contribution

The paper presents a novel distributed PCA algorithm that simplifies network architecture and training by removing intralayer weights.

Findings

Successfully computes PCA in a distributed manner

Reduces number of trained weights by half

Proves convergence and effectiveness of the method

Abstract

In this paper we propose an algorithm, Simple Hebbian PCA, and prove that it is able to calculate the principal component analysis (PCA) in a distributed fashion across nodes. It simplifies existing network structures by removing intralayer weights, essentially cutting the number of weights that need to be trained in half.