Minas Pertselakis

A Multi-clustering Fusion Algorithm

A multi-clustering fusion method is presented based on combining several runs of a clustering algorithm resulting in a common partition. More specifically, the results of several independent runs of the same clustering algorithm are appropriately combined to obtain a partition of the data which is not affected by initialization and overcomes the instabilities of clustering methods. Finally, the fusion procedure starts with the clusters produced by the combining part and finds the optimal number of clusters in the data set according to some predefined criteria. The unsupervised multi-clustering method implemented in this work is quite general. There is ample room for the implementation and testing with any existing clustering algorithm that has unstable results. Experiments using both simulated and real data sets indicate that the multi-clustering fusion algorithm is able to partition a set of data points to the optimal number of clusters not constrained to be hyperspherically shaped.

An adaptable Gaussian neuro-fuzzy classifier

The concept of semantic and context aware intelligent systems provides a vision for the Information Society where the emphasis lays on computing applications that can sense context from the people and the environment and wrap that knowledge into adaptable behavior. In this framework the proper and automatic classification of data gathered by sensors is of major importance. Our approach describes a model that operates as a self-evaluating classifier using on-line re-clustering, addressing adequately the basic issues of modern demands. The novelty of the model lies in a flexible and efficient initialization technique that first partitions the data space utilizing Gaussian distributions and then merges clusters so as to produce an effective partitioning.

A SOM-based classifier with enhanced structure learning

This work introduces an innovative synergistic model that aims to improve the efficiency of a neuro-fuzzy classifier, providing the means of online adaptation and fast learning. It combines the advantages of a self-organized map (SOM) network, as well as the benefits of a structure allocation fuzzy neural network. The system initializes its parameters using the clustering result on the SOM structure, while a novel approach of evaluating the input features leads to a more efficient way of handling the on-line learning rate of the training process. Experimental results on benchmark classification problems showed that this robust combination can also tackle tasks of great dimensionality in a successful manner.

Adaptive Rule-Based Facial Expression Recognition

This paper addresses the problem of emotion recognition in faces through an intelligent neuro- fuzzy system, which is capable of analysing facial features extracted following the MPEG-4 standard and classifying facial images according to the underlying emotional states, following rules derived from expression profiles. Results are presented which illustrate the capability of the developed system to analyse and recognise facial expressions in man-machine interaction applications.

WIRELESS DISTRIBUTED IMPLEMENTATION OF A FUZZY NEURAL CLASSIFICATION SYSTEM

Recent years have seen a surge of interest in the field of pervasive context-aware computing. In this framework we propose a novel real implementation of an adaptive self-configurable system, applied within the scope of wireless ad-hoc networks. WiDFuNC is an integrated system that consists of an intelligent unit implemented on a real PDA, a number of sensors and a remote server device to form an efficient prototype system that can be applied in various domains. This implementation of WiDFuNC focuses on pure classification problems with satisfactory experimental results, presenting great adaptability and context-awareness.

The RAPS Process Model: A framework for the identification and the management of risks against patient safety (RAPS)

ReMINE is a novel framework architecture for the management of risk against patient safety (RAPS) in health care systems. ReMINE allows for the collection and analysis of RAPS-related data through a semantic approach which offers a fast and secure extraction of data and correlation of the information across several domains. In this respect, the ReMINE platform will promote an early RAPS detection and forecast that will assist risk managers to provide reliable solutions. The core mechanism to control and handle all the knowledge and data from the various system components is called RAPS Process Model. This mechanism interacts also with the user by receiving and transmitting information using business process modeling and ontological engineering methodologies.

An Adaptive Resource Allocating Neuro-Fuzzy Inference System with Sensitivity Analysis Resource Control

Adaptability in non-stationary contexts is a very important property and a constant desire for modern intelligent systems and is usually associated with dynamic system behaviors. In this framework, we present a novel methodology of dynamic resource control and optimization for neurofuzzy inference systems. Our approach involves a neurofuzzy model with structural learning capabilities that adds rule nodes when necessary during the training phase. Sensitivity analysis is then applied to the trained network so as to evaluate the network rules and control their usage in a dynamic manner based on a confidence threshold. Therefore, on one hand, we result in a well-balanced structure with an improved adaptive behavior and, on the other hand, we propose a way to control and restrict the “curse of dimensionality”. The experimental results on a number of classification problems prove clearly the strengths and benefits of this approach.

Exploiting Decision Trees in Product-based Fuzzy Neural Modeling to Generate Rules with Dynamically Reduced Dimensionality

Decision trees are commonly employed as data classifiers in various research fields, but also in real-world application domains. In the fuzzy neural framework, decision trees can offer valuable assistance in determining a proper initial system structure, which means not only feature selection, but also rule extraction and organization. This paper proposes a synergistic model that combines the advantages of a subsethood-product neural fuzzy inference system and a CART algorithm, in order to create a novel architecture and generate fuzzy rules of the form “IF - THEN IF”, where the first “IF” concerns the primary attributes and the second “IF” the secondary attributes of the given dataset as defined by our method. The resulted structure eliminates certain drawbacks of both techniques and produces a compact, comprehensible and efficient rulebase. Experiments in benchmark classification tasks prove that this method does not only reduce computational cost, but it also maintains performance at high levels, offering fast and accurate processing during realtime operations.

Dynamic Rule Structuring based on Rule-Activation Sensitivity Analysis

In the field of soft computing the need of more flexible, fast and robust approaches is evident. The main idea behind this paper is the construction of sets of fuzzy rules with a hierarchical structure derived from data. The appropriate set is then applied on demand. Instead of rule pruning or rule refinement, we propose a way that sorts the total of rules of a well-trained fuzzy neural network in order of significance and creates prioritized rule sets using sensitivity analysis. A confidence measure indicates the appropriate set to be utilized at any given time. Experiments in benchmark classification tasks prove that this method does not only reduce computational cost, but it also maintains performance at the same levels, offering fast processing during real-time operations