Work

Margin-Based Training of HDC Classifiers

Year: 2025

Type: article

Abstract: The explicit kernel transformation of input data vectors to their distributed high-dimensional representations has recently been receiving increasing attention in the field of hyperdimensional computing (HDC). The main argument is that such representations endow simpler last-leg classification models, often referred to as HDC classifiers. HDC models have obvious advantages over resource-intensive deep learning models for use cases requiring fast, energy-efficient computations both for model training and deploying. Recent approaches to training HDC classifiers have primarily focused on various methods for selecting individual learning rates for incorrectly classified samples. In contrast to these methods, we propose an alternative strategy where the decision to learn is based on a margin applied to the classifier scores. This approach ensures that even correctly classified samples within the specified margin are utilized in training the model. This leads to improved test performances while maintaining a basic learning rule with a fixed (unit) learning rate. We propose and empirically evaluate two such strategies, incorporating either an additive or multiplicative margin, on the standard subset of the UCI collection, consisting of 121 datasets. Our approach demonstrates superior mean accuracy compared to other HDC classifiers with iterative error-correcting training.

Source: Big Data and Cognitive Computing

DOI: 10.3390/bdcc9030068

Authors: Laura Smets, Dmitri Rachkovskij, Evgeny Osipov, Werner Van Leekwijck, Olexander Volkov, Steven Latré

Institutions: University of Antwerp, IMEC, Luleå University of Technology

Language: English

Topic: Machine Learning and ELM

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $2,172

Funders: National Research Foundation of Ukraine, Stiftelsen för Strategisk Forskning, Vetenskapsrådet