Advances in neural information processing systems 2
Knowledge-based artificial neural networks
Artificial Intelligence
Formal Concept Analysis: Mathematical Foundations
Formal Concept Analysis: Mathematical Foundations
Computing iceberg concept lattices with TITANIC
Data & Knowledge Engineering
Mining Minimal Non-redundant Association Rules Using Frequent Closed Itemsets
CL '00 Proceedings of the First International Conference on Computational Logic
Data Mining: Concepts and Techniques
Data Mining: Concepts and Techniques
Data Mining: Practical Machine Learning Tools and Techniques, Second Edition (Morgan Kaufmann Series in Data Management Systems)
IGB: a new informative generic base of association rules
PAKDD'05 Proceedings of the 9th Pacific-Asia conference on Advances in Knowledge Discovery and Data Mining
Review: Formal Concept Analysis in knowledge processing: A survey on models and techniques
Expert Systems with Applications: An International Journal
| Hi-index | 0.00 |
Multi-layer neural networks have been successfully applied in a wide range of supervised and unsupervised learning applications. However defining its architecture is a difficult task, and might make their usage very complicated. To solve this problem, a rule-based model, KBANN, was previously introduced making use of an apriori knowledge to build the network architecture. Neithertheless this apriori knowledge is not always available when dealing with real world applications. Other methods presented in the literature propose to find directly the neural network architecture by incrementally adding new hidden neurons (or layers) to the existing network, network which initially has no hidden layer. Recently, a novel neural network approach CLANN based on concept lattices was proposed with the advantage to be suitable for finding the architecture of the neural network when the apriori knowledge is not available. However CLANN is limited to application with only two-class data, which is not often the case in practice. In this paper we propose a novel approach M-CLANN in order to treat multi-class data. Carried out experiments showed the soundness and efficiency of our approach on different UCI datasets compared to standard machine learning systems. It also comes out that M-CLANN model considerably improved CLANN model when dealing with two-class data.