2024-03-29T11:33:12Z
https://meral.edu.mm/oai
oai:meral.edu.mm:recid/4479
2022-03-24T23:16:18Z
1582963302567:1597824273898
user-ucsy
RBF Neural Network Based on Clonal Selection Algorithm for Medical Data Diagnosis
Thandar, Aye Mya
Khine, May Aye
In artificial neural networks, the parametersmay include the number of layers, the number ofhidden units, the activation function and the algorithmparameters such as learning rate for optimization.Many researchers have proven that the training ofartificial neural networks is a complex process andmethods of training are highly varied. Some attemptto approximate the process of biological neurons butmany diverge greatly from them in an attempt to findmore computationally efficient methods to achieveoptimal or near-optimal weights. Although radialbasis function networks (RBF) are well known forrequiring short training period among artificialneural networks, these methods perform a localsearch and they can easily fall in local minima byproducing sub-optimal solutions. Therefore, theperformance of network training is not good and theaccuracy is low for RBF neural networks. Thetraditional network weight training generally usesgradient descent method and it can not get the globaloptimum. Training the weights by optimizationmethod can find the weight set that approaches globaloptimum while do not need to compute gradientinformation and it can help to reduce error rate innetwork training .Clonal selection algorithm is aglobal search among optimization method and it canprovide an efficient alternative for the optimization ofneural networks. In this paper, we use clonal selectionalgorithm to adjust weight units which are importantto improve network training in RBF neural network.
2012-02-28
http://hdl.handle.net/20.500.12678/0000004479
https://meral.edu.mm/records/4479