On Effects of Compression with Hyperdimensional Computing in Distributed Randomized Neural Networks

TL;DR

This paper explores how different compression techniques affect distributed randomized neural networks based on hyperdimensional computing, aiming to improve communication efficiency and flexibility.

Contribution

It introduces a more flexible compression approach for distributed hyperdimensional neural networks and compares it with existing methods.

Findings

Flexible compression improves communication efficiency

Comparison shows advantages over traditional methods

Enhanced adaptability in distributed learning systems

Abstract

A change of the prevalent supervised learning techniques is foreseeable in the near future: from the complex, computational expensive algorithms to more flexible and elementary training ones. The strong revitalization of randomized algorithms can be framed in this prospect steering. We recently proposed a model for distributed classification based on randomized neural networks and hyperdimensional computing, which takes into account cost of information exchange between agents using compression. The use of compression is important as it addresses the issues related to the communication bottleneck, however, the original approach is rigid in the way the compression is used. Therefore, in this work, we propose a more flexible approach to compression and compare it to conventional compression algorithms, dimensionality reduction, and quantization techniques.