Symbolic interpretation of artificial neural networks using genetic algorithms
Symbolic interpretation of artificial neural networks using genetic algorithms
The knowledge acquired during the learning of artificial neural networks (ANNs) is coded as values in synapticweights, which makes their interpretations difficult, hence the name of the black box. The aim of this work is to providea comprehensible interpretation of the ANN’s decisions by extracting symbolic rules. We improve the performance of ourextraction algorithm by combining the ANN with a genetic algorithm. Misleading rules whose support and confidencevalues are less than fixed thresholds are removed and, as a result, the comprehensibility is improved. The extracted rulesare evaluated and compared with other works. The results show good performance of our proposal in terms of fidelityand accuracy
___
- Pandya MD, Patel JR. A survey: artificial neural network for character recognition. In: Proceedings of Fifth
International Conference on Soft Computing for Problem Solving; 2016; Singapore: Springer. pp. 403-410.
- Hayashi Y, Setiono R, Azcarraga A. Neural network training and rule extraction with augmented discretized input.
Neurocomputing 2016; 207: 610-622.
- Augasta M, Kathirvalavakumar T. Reverse engineering the neural networks for rule extraction in classification
problems. Neural Process Lett 2012; 35: 131-150.
- Craven MW. Extracting Comprehensible Models from Trained Neural Networks. PhD, University of Wisconsin,
Madison, WI, USA, 1996.
- Markowska-Kaczmar U. Evolutionary approaches to rule extraction from neural networks. Stud Comput Intell 2008;
82: 177-209.
- Craven M, Shavlik J. Extracting tree-structured representations of trained networks. Adv Neur Info Proc Sys 1996;
8: 24-30.
- Etemadi H, Ahmadpour A, Moshashaei SM. Earnings per share forecast using extracted rules from trained neural
network by genetic algorithm. Comput Econ 2015; 46: 55-63.
- Yedjour D, Benyettou A, Yedjour H. Combining Quine Mc-Cluskey and genetic algorithms for extracting rules from
trained neural networks. Asian J Appl Sci 2011; 4: 72-80.
- Towell G, Shavlik JW. The extraction of refined rules from knowledge-based neural networks. Mach Learn 1993;
131: 71-101.
- Taha I, Ghosh J. Symbolic interpretation of artificial neural networks. IEEE T Knowl Data Eng 1999; 11: 448-463.
- Andrews R, Diederich J, Aickle AB. Survey and critique of techniques for extracting rules from trained artificial
neural networks. Knowl Based Syst 1995; 8: 373-389.
- Santos R, Nievola J, Freitas A. Extracting comprehensible rules from neural networks via genetic algorithm. In:
Proceedings of the IEEE Symposium on Combination of Evolutionary Algorithm and Neural Network; San Antonio,
TX, USA. 2000, pp. 130-139.
- Dhanalaxmi B, Naidu GA, Anuradha K. Adaptive PSO based association rule mining technique for software defect
classification using ANN. Procedia Comput Sci 2015; 46: 432-442.
- Blake CC, Merz C. UCI Repository of Machine Learning Databases. University of California, Irvine, Dept. of
Information and Computer Sciences 1998.
- Setiono R, Liu H. Understanding neural networks via rule extraction. In: Proceedings of the 14th International
Joint Conference on Artificial Intelligence (IJCAI); 1995. pp. 480-485.
- Saad EW, Wunsch DC. Neural network explanation using inversion. Neural Networks 2007; 20: 78-93.
- Setiono R, Liu H. NeuroLinear: From neural networks to oblique decision rules. Neurocomputing 1997; 17: 1-24.
- Kim D, Lee J. Handling continuous-valued attributes in decision tree with neural network modeling. In: Proceedings
of the 11th European Conference on Machine Learning; 31 May–2 June 2000; Barcelona, Catalonia, Spain. Berlin,
Germany: Springer, pp. 211-219.