Implementation of hop field neural network model in the area of associative memory

Kumar, Jagdish and Kalra, Sandeep and Chopra, A.K. (2002) Implementation of hop field neural network model in the area of associative memory. Journal of Instrument Society of India, 32 (4). pp. 296-301.

[img] PDF
Restricted to Registered users only

Download (193Kb)
Official URL:


The human memory has its ability to learn about many new things without necessarily forgetting them, learnt in the past. Exactly the same can be thought about the Artificial Neural Networks system, How can its learning system remain responsive in response to significant input, yet remain stable in response to irrelevant input? How does the system retain previously learnt information while continuing to learn about new information. This paper attempts to address the theory and implementation of Hopfield Neural Network in the area of Associative Memory(AM). There are three types of Associative Memory: Hetero-associative Memory, Interpolative Memory and Auto-associative Memory. Human memory works on the basis of partial knowledge of its contents and or association with other information. This may be called as “Content-Addressible Memory”. We provide a small part of input pattern to the system, the net or by self excitation the whole pattern will be regenerated. Neural network simulations have allowed researchers to study how and under what conditions such studies are that associative memory can do better in the presence of a certain level of internal noise or with certain level of forgetfulness, The system, distributed as one memory is distributed over many synapses, superimposed because one synapse can be involved in several memories and robust because altering a few synapses degrade the performance by very little.

Item Type: Article
Subjects: CSIO > Medical Instrumentation
Depositing User: Ms. J Shrivastav
Date Deposited: 09 Mar 2012 16:34
Last Modified: 23 Apr 2012 10:04

Actions (login required)

View Item View Item