Black-Box Classifier Interpretation Using Decision Tree and Fuzzy Logic-Based Classifier Implementation

• DOI: 10.5391/ijfis.2016.16.1.27
• Publication Date: 2016-04-26T04:37:26Z
• Authors: Hansoo Lee; Sungshin Kim
• Keywords: Quadratic classifier; Margin classifier

Black-box classifiers, such as artificial neural network and support vector machine, are a popular classifier because of its remarkable performance.They applied in various fields inductive inferences, classifications, or regressions.However, by characteristics, they cannot provide appropriate explanations how the classification results derived.Therefore, there plenty actively discussed researches about interpreting trained black-box classifiers.In this paper, we propose method to make fuzzy logic-based using extracted rules from machine order interpret internal structures.As an object classification, anomalous propagation echo is selected which occurs frequently radar data becomes problem precipitation estimation process.After applying clustering method, learning dataset generated clusters.Using dataset, implemented.After that, decision trees for each generated.And used implement simplified classifiers rule extraction input selection.Finally, can verify compare performances.With actual occurrence cased echo, determine inner structures classifiers.