Answers are grounded in this paper and cite their sources.
Demonstrates the decomposition of cyclic-group-ring-valued Hopfield networks into simpler network types, implying enhanced learning algorithms.
Hopfield networks have been extended to multistate Hopfield networks, since complex-valued Hopfield networks (CVHNs) were proposed. Multistate Hopfield networks have often been applied to storage of multilevel data, such as images. Cyclic-group-ring-valued Hopfield networks (CGRVHNs) are multistate Hopfield networks proposed in the framework of group-ring-valued Hopfield networks. In this paper, we prove that a CGRVHN decomposes to a CVHN and a hyperbolic-valued Hopfield network (HVHN), where the decomposition is not unique. Therefore, the CGRVHNs are equivalent to mixed Hopfield networks of CVHNs and HVHNs. Moreover, we introduce hebbian rule for CGRVHNs, and prove that it also decomposes to those for CVHNs and HVHNs. These decompositions are expected to lead to improvement in the learning algorithms and analysis of the inherent properties.
No takes yet. Share an insight, caveat, or question.
Masaki Kobayashi (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: