Neuro-Fuzzy Decision Tree by Fuzzy ID3 Algorithm and Its Application to Anti-Dumping Early-Warning System

Fuzzy decision trees (FDT) are one of the most popular choices for learning and reasoning from dataset. They have undergone a number of alterations to language and measurement uncertainties. However, they are poor in classification accuracy. In this paper, we proposed neuro-fuzzy decision tree. Neuro-fuzzy decision tree (a fuzzy decision tree structure with neural like parameter adaptation strategy) improves FDT´s classification accuracy and extracts more accuracy human interpretable classification rules. In the forward cycle, we construct fuzzy decision trees using fuzzy ID3. In the feedback cycle, parameters of fuzzy decision trees have been adapted using stochastic gradient descent algorithm by traversing back from leaf to root nodes. In this paper, we proposed a new anti-dumping early-warning system. The early-warning system based on neuro-fuzzy decision tree modeling method is different from traditional modeling methods. The other new attempt is the setting of early-warning intervals. The result of the positive research indicated that this system is very valid for anti-dumping prediction and it will have a good application prospect in this area