Learnable Reasoning Rules for Neuro-Vector-Symbolic Architectures via Hidden Markov Models

20260 citationsPreprintgreen Open Access

Authors

Nicolas Cloutier · Laboratory of Excellence GR-Ex

Abstract

We propose HMM-VSA, a learnable reasoning rule frameworkwhere Hidden Markov Models operate directly on block-codehypervector representations, with Graph Finite-State Automaton(GFSA) theory providing the formal foundation.

Topics & Keywords

Advanced Graph Neural Networks Graph Theory and Algorithms Model-Driven Software Engineering Techniques

Publication Details

Published in: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.19190708

Command Palette

Learnable Reasoning Rules for Neuro-Vector-Symbolic Architectures via Hidden Markov Models

Authors

Abstract

Topics & Keywords

Publication Details