Associative Memory with Small-World Adaptive Structure through Annealed Rewiring

摘要

A novel associative memory network based on small-world adaptive structure through annealed rewiring is proposed in this paper. Aimed at overcoming the disadvantage of quenched rewiring as random short-cuts formation of the existing methods, this new model takes the ideology of Harmonious Unifying Hybrid Preferential Model into account, and investigates the optimal synaptic dilution strategy under the constraint of limited amount of synapses. Based on the theoretical analysis above, the new model breaks the traditional manner of quenched rewiring (short-cuts formation) but instead constructs a task-dependent network structure through annealed way which is much closer to human brain as possessing small-world architecture and can also achieve better performance than the existing counterparts of the same class. The rationality and validity of the proposed model is validated from great number of experiments.