This paper proposes an implementation scheme of K-class classification problem using systems of multiple neural networks. Usually, a multi-class problem is decomposed into simple sub-problems solved independently using similar single neural networks. For the reason that these sub-problems are not equivalent in their complexity, we propose a system that includes reinforced networks destined to solve complicated parts of the entire problem. Our approach is inspired from principles of the multi-classifiers systems and the labeled classification, which aims to improve performances of the networks trained by the Back-Propagation algorithm. We propose two implementation schemes based on both OAO (one-against-all) and OAA (one-against-one). The proposed models are evaluated using iris and human thigh databases.
In knowledge-based systems, uncertainty in propositions can be represented by various degrees of belief encoded by numerical or symbolic values. The use of symbolic values is necessary in areas where the exact numerical values associated with a fact are unknown by experts. In this paper we present an expert system of supervised automatic classification based on a symbolic approach. This last is composed of two sub-systems. The first sub-system automatically generates the production rules using training set; the generated rules are accompanied by a symbolic degree of belief which characterizes their classes of memberships. The second is the inference system, which receives in entry the base of rules and the object to classify. Using classical reasoning (Modus Ponens), the inference system provides the membership class of this object with a certain symbolic degree of belief. Methods to evaluate the degree of belief are numerous but they are often tarnished with uncertainty. To appreciate the performances of our symbolic approach, tests are made on the Iris data basis.
Setting a compact and accurate rule base constitutes the principal objective in designing fuzzy rule-based classifiers. In this regard, the authors propose a designing scheme based on the combination of the subtractive clustering (SC) and the particle swarm optimization (PSO). The main idea relies on the application of the SC on each class separately and with a different radius in order to generate regions that are more accurate, and to represent each region by a fuzzy rule. However, the number of rules is then affected by the radiuses, which are the main preset parameters of the SC. The PSO is therefore used to define the optimal radiuses. To get good compromise accuracy-compactness, the authors propose using a multi-objective function for the PSO. The performances of the proposed method are tested on well-known data sets and compared with several state-of-the-art methods.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.