Knowledge discovery in databases: an overview
AI Magazine
Rule extraction from trained neural networks using genetic algorithms
Proceedings of the second world congress on Nonlinear analysts: part 3
Data mining: concepts and techniques
Data mining: concepts and techniques
The Influence of Parameters in Evolutionary Based Rule Extraction Method from Neural Network
ISDA '05 Proceedings of the 5th International Conference on Intelligent Systems Design and Applications
A Dual-Objective Evolutionary Algorithm for Rules Extraction in Data Mining
Computational Optimization and Applications
IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
A new approach to classification based on association rule mining
Decision Support Systems
A greedy classification algorithm based on association rule
Applied Soft Computing
TACO-miner: An ant colony based algorithm for rule extraction from trained neural networks
Expert Systems with Applications: An International Journal
Engineering Applications of Artificial Intelligence
Optimizing the modified fuzzy ant-miner for efficient medical diagnosis
Applied Intelligence
Hi-index | 0.00 |
A common problem in Data Mining (DM) is the presence of noise in the data being mined. Artificial neural networks (ANN) are robust and have a good tolerance to noise, which makes them suitable for mining very noisy data. Although they may achieve high classification accuracy, they have the well-known disadvantage of having black-box nature and not discovering any high-level rule that can be used as a support for human understanding. The main challenge in using ANN in DM applications is to get explicit knowledge from these models. For this purpose, a study on knowledge acquirement from trained ANNs for classification problems is presented. The proposed method uses Touring Ant Colony Optimization (TACO) algorithm for extracting accurate and comprehensible rules from databases via trained artificial neural networks. The suggested algorithm is experimentally evaluated on different benchmark data sets. Results show that the proposed approach has a potential to generate accurate and concise rules.