Chromosome Classification Using A Multi-layer Perceptron Neural Net
Qi Wu, Paul L. Suetens, A. Oosterlinck · 2005
In this paper we present our recent study on using neural net systems for automated classification of human chromosomes. A multi-layer perceptron classifier was implemented and trained by a back-propagation algorithm. In comparison with traditional statistical pattern classification techniques, the neural net approach exhibits potential benefits of superior performance, adaptive learning capabilities, and high computation rates provided by massive parallelism. Results of the experiments carried out using both the perceptron and a Bayes classifier on a data set of chromosome feature measurements are given and compared.