A New Unsupervised/Reinforcement Learning Method In Spiking Pattern Classification Networks

Research Square (Research Square), Journal Year: 2023, Volume and Issue: unknown

Published: Nov. 8, 2023

Computations adapted from the interactions of neurons in nervous system may be a capable platform that can create powerful machines terms cognitive abilities such as real-time learning, decision-making and generalization. In this regard, here an intelligent machine based on basic approved mechanisms has been proposed. Therefore, input layer presented is retinal model middle output composed population pyramidal neurons/ interneurons, AMPA/GABA receptors, excitatory/inhibitory neurotransmitters. A bio-adapted structure requires learning biological evidence. Similarly, new mechanism unsupervised (Power-STDP) reinforcement procedure (Actor-Critic algorithm) was proposed which called PSAC algorithm. Three challenging datasets MNIST, EMNIST, CIFAR10 were used to confirm performance algorithm compared deep spiking networks, respectively accuracies 97.7%, 97.95% (digits) 93.73% (letters), 93.6% have obtained, shows improvement accuracy previous networks. addition being more accurate than spike-based methods, approach higher convergence speed training process. Although obtained classification are slightly lower but speed, low power consumption if implemented neuromorphic platforms, advantages network.

Language: Английский

Analysis of the Effectiveness of Integrating Traditional Chinese Medicine Culture and Tang Grass Pattern into Medicinal Packaging Design for User Experience Enhancement - Simulation and Modeling based on SNNs Neural Network FittingIntegrating Traditional Chinese Medicine Culture and Tang Grass Pattern into Medicinal Packaging Design for User Experience Enhancement

Published: Oct. 20, 2023

[Background] This study investigates the effectiveness of integrating traditional Chinese medicine culture and Tang grass pattern into medicinal packaging design for enhancing user experience. [Method]The research addresses complex data processing large-scale model challenges associated with this topic. An improved dual machine learning causal inference is proposed, based on SNNs network structure incorporating a multi-strategy optimization framework. The achieves enhanced accuracy while reducing size computational requirements. [Result]Experimental results demonstrate that model, compared to previous exhibits reduced workload, prediction 96.2%. [Implication]The algorithm provides better evaluating

Language: Английский