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

Key Points

• To develop a framework that enhances reasoning rules using Hidden Markov Models in vector-symbolic architectures.
• Developed HMM-VSA framework integrating Hidden Markov Models with hypervector representations.
• Utilized Graph Finite-State Automaton theory for formal foundational support.
• Demonstrates the feasibility of HMM-VSA for reasoning tasks.
• Indicates potential improvements in processing and reasoning efficiency.