Main Article Content
Abstract
Neural network is very accurate in results that’s why it is mainly used for decision making but because of its black box nature we cannot see its internal working. So, rule extraction is a way to solve this issue. This could be seen as the root for rule extraction investigation in neural network. To describe the inference process, we need to follow a routine to take out rules from a neural network. Three commonly algorithms used for extracting rules from a simple network. This includes decompositional approach, pedagogical approach, eclectic methods. The main criticism of neural networks is that it is difficult to understand the decision-making process in neural networks. One of the most widely considered limitations of ANNs is their inability to explain. Nonetheless, the basis of decisions must be considered, as this computer support systems are often used in decision-making applications like medical diagnosis. The best way to represent extracted knowledge from neural networks is by using IF THEN rules. To describe a neural network, IF-THEN rule is the best way to eliminate decision tree from a network. In this paper, I have discussed three extraction algorithms and also about how decision tree is helping in reaching to particular decision.