S B Bagal and U V Kulkarni. Article: Genetic Algorithm based Rule Extraction from Pruned Modified Fuzzy Hyperline Segment Neural Network for Pattern Classification. IJCA Proceedings on International Conference on Internet of Things, Next Generation Networks and Cloud Computing ICINC 2016(2):25-33, July 2016. Full text available. BibTeX
@article{key:article, author = {S. B. Bagal and U. V. Kulkarni}, title = {Article: Genetic Algorithm based Rule Extraction from Pruned Modified Fuzzy Hyperline Segment Neural Network for Pattern Classification}, journal = {IJCA Proceedings on International Conference on Internet of Things, Next Generation Networks and Cloud Computing}, year = {2016}, volume = {ICINC 2016}, number = {2}, pages = {25-33}, month = {July}, note = {Full text available} }
The Pruned modified fuzzy hyperline segment neural network (PMFHLSNN) is pruned extension of Fuzzy hyperline segment neural network (FHLSNN) with modification in the testing phase. In this paper, a genetic algorithm based rule extractor (GA-PMFHLSNN) is proposed to extract a small set of compact and comprehensible fuzzy if-then rules with high classification accuracy from the PMFHLSNN. After pruning, open hyperline segments are generated from the remaining hyperline segments and a "don't care" approach is adopted by GA rule extractor to minimize the number of features in the extracted rules with higher classification accuracy. The performance of FHLSNN, PMFHLSNN and GA-PMFHLSNN are evaluated using tenfold cross-validation for five benchmark problems and handwritten character database. All the results show that the proposed approach can extract a set of compact and comprehensible rules with high classification accuracy for all the selected datasets.