Methodology for Knowledge Extraction from Trained Artificial Neural Networks

2018
Andrejs Bondarenko, Ludmila Aleksejeva

Artificial neural networks are widely spread models that outperform more basic, but explainable machine learning models like classification decision tree. However, their lack of explainability severely limits their area of application. All mission critical areas or law regulated areas (like European GDPR) require model to be explained. Explainability allows model validation for correctness and lack of bias. Thus, methods for knowledge extraction from artificial neural networks have gained attention and development efforts. The present paper addresses this problem and describes a knowledge extraction methodology which can be applied to classification problems. It is based on previous research and allows knowledge to be extracted from trained fully connected feed-forward artificial neural network, from radial basis function neural network and from hyper-polytope based classifier in the form of binary classification decision tree, elliptical rules and If−Then rules.

Atslēgas vārdi
Artificial neural network, feed-forward neural networks, knowledge acquisition, knowledge extraction, radial basis function neural networks
DOI
10.7250/itms-2018-0001
Hipersaite
https://itms-journals.rtu.lv/article/view/itms-2018-0001

Bondarenko, A., Aleksejeva, L. Methodology for Knowledge Extraction from Trained Artificial Neural Networks. Information Technology and Management Science, 2018, Vol. 21, No. 1, 6.-14.lpp. ISSN 2255-9086. e-ISSN 2255-9094. Pieejams: doi:10.7250/itms-2018-0001

Publikācijas valoda
English (en)