返回Google圖書搜尋

Learning in Neural Networks

其他書名	A Comparison of Networks of Hodgkin-Huxley Neurons and Networks of Map Model Neurons Each Endowed with Spike-time-dependent Plasticity
出版	Ohio State University, 2007
URL	http://books.google.com.hk/books?id=MO3dXwAACAAJ&hl=&source=gbs_api

註釋Abstract: This paper investigates the effect of adding adjustable synapses to a network of Hodgkin-Huxley neurons. The synapse-learning rule used is multiplicative spike-time- dependent plasticity. Using the computer program, LearningNets, this project investigates the resulting firing patterns after a network of forty neurons is allowed to run for 4000 milliseconds. All synapses are allowed to evolve for this whole time period. Half of the neurons are biased with a sinusoidal applied current for the first three seconds. Results include investigating synchronization of neurons, the effect of the number of synapses allowed, the effect of the number of inhibitory synapses on the resulting firing of the neurons, the effect of varying the time delay between neurons, the distribution of synapse weights, and the number of neurons which are active at a given time. All of the results are compared to the same studies for a network of 40 map model neurons with additive spike-time-dependent plasticity. This study finds that both models lead to learning when the right parameters are chosen. The Hodgkin-Huxley model leads to a more biological network with higher levels of synchronization.