Antanas Andrius Bielskis

An approach of multi-agent control of bio-robots using intelligent recognition diagnosis of persons with moving disabilities

Abstract The aims of this research are focused on the construction of intellectualized equipments for people with moving disabilities to help them in sustainable integration into environment. The problem is to reveal main components of diagnosis of disabled persons, as well as to develop decision making models which are integrated into the control mechanisms of the special equipments, that are assigned to the class of bio‐robots. This paper analyses the approach of the construction of such type of bio‐robots with possibilities to integrate different knowledge representation techniques for the development of the reinforcement framework with multiple cooperative agents for the recognition of the diagnosis of emotional situation of disabled persons. Large‐scale of multidimensional recognitions of emotional diagnosis of disabled persons often generate a large amount of multi‐dimensional data with complex recognition mechanisms, based on the integration of different knowledge representation techniques and comp...

Classifiers fusion in recognition of wheat varieties

Five wheat varieties (Bezostaja, Cesit1252, Daǧdas, Gerek, Kiziltan traded in Konya Exchange of Commerce, Turkey), characterized by nine geometric and three colour descriptive features have been classified by multiple classier system where pair-wise SLP or SV classifiers served as base experts. In addition to standard voting and Hastie and Tibshirani fusion rules, two new ones were suggested that allowed reducing the generalization error up to 5%. In classifying of kernel lots, we may obtain faultless grain recognition.

Data mining approaches for intelligent E-Social care decision support system

Large-scale of multidimensional recognitions of emotional diagnosis of disabled persons often generate large amount of multidimensional data with complex recognition mechanisms. The problem is to reveal main components of diagnosis and to construct flexible decision making support system. Sensors can easily record primary data, however the recognition of abnormal situations, clusterization of emotional stages and resolution for certain type of diagnosis is oncoming issue for bio-robot constructors. This paper analyses the possibilities of integration of different knowledge representation techniques, especially data mining methods, for development of the reinforcement framework with multiple cooperative agents for recognition of the prediction criteria of diagnosis of emotional situation of disabled persons. The research results present further development of model of framework with integration of the evaluation of data mining methods for wheelchair type robots working in real time by providing movement support for disabled individuals.

An Intelligent Control System Based on Non-Invasive Man Machine Interaction

This paper presents further development of intelligent multi-agent based e-health care system for people with movement disabilities. The research results present further development of multi-layered model of this system with integration of fuzzy neural control of speed of two wheelchair type robots working in real time by providing movement support for disabled individuals. An approach of filtering of skin conductance (SC) signals using Nadaraya-Watson kernel regression smoothing for emotion recognition of disabled individuals is described and implemented in the system by R software tool. The unsupervised clustering by self organizing maps (SOM) of data sample of physiological parameters extracted from SC signals was proposed in order to reduce teacher noise as well as to increase of speed and accuracy of learning process of multi-layer perceptron (MLP) training.

A pool of classifiers by SLP: a multi-class case

Dynamics of training the group of single layer perceptrons aimed to solve multi-class pattern recognition problem is studied. It is shown that in special training of the perceptrons, one may obtain a pool of different classification algorithms. Means to improve training speed and reduce generalization error are studied. Training dynamics is illustrated by solving artificial multi-class pattern recognition task and important real world problem: detection of ten types of yeast infections from 1500 spectral features.