Maximization of the information divergence from multinomial distributions
The Relational Equations approach is one of the most usual ones for describing (Fuzzy) Systems and in most cases, it is the final expression for other descriptions. This is why the identification of Relational Equations from a set of examples has received considerable atention in the specialized literature. This paper is devoted to this topic, more specifically to the topic of max-min neural networks for identification. Three methods of learning Fuzzy Systems are developed by combining the most...
Recently a new interesting architecture of neural networks called “mixture of experts” has been proposed as a tool of real multivariate approximation or prediction. We show that the underlying problem is closely related to approximating the joint probability density of involved variables by finite mixture. Particularly, assuming normal mixtures, we can explicitly write the conditional expectation formula which can be interpreted as a mixture-of- experts network. In this way the related optimization...
A framework for multi-label classification extended by Error Correcting Output Codes (ECOCs) is introduced and empirically examined in the article. The solution assumes the base multi-label classifiers to be a noisy channel and applies ECOCs in order to recover the classification errors made by individual classifiers. The framework was examined through exhaustive studies over combinations of three distinct classification algorithms and four ECOC methods employed in the multi-label classification...
Multiple-Instance Learning (MIL) has attracted much attention of the machine learning community in recent years and many real-world applications have been successfully formulated as MIL problems. Over the past few years, several Instance Selection-based MIL (ISMIL) algorithms have been presented by using the concept of the embedding space. Although they delivered very promising performance, they often require long computation times for instance selection, leading to a low efficiency of the whole...
Genetic algorithms (GAs) represent a class of adaptive search techniques inspired by natural evolution mechanisms. The search properties of GAs make them suitable to be used in machine learning processes and for developing fuzzy systems, the so-called genetic fuzzy systems (GFSs). In this contribution, we discuss genetics-based machine learning processes presenting the iterative rule learning approach, and a special kind of GFS, a multi-stage GFS based on the iterative rule learning approach, by...