Edward Puchala

A bayes algorithm for the multitask pattern recognition problem - direct approach

The paper presents algorithms of the multitask recognition for the direct approach. First one, with full probabilistic information and second one, algorithms with learning sequence. Algorithm with full probabilistic information was working on basis of Bayes decision theory. Full probabilistic information in a pattern recognition task, denotes a knowledge of the classes probabilities and the class-conditional probability density functions. Optimal algorithm for the selected loss function will be presented. Some tests for algorithm with learning were done.

Hybrid Pattern Recognition Algorithms with the Statistical Model Applied to the Computer-Aided Medical Diagnosis

The present paper is devoted to the pattern recognition procedures that simultaneously use the information contained in the empirical data (learning set) and the set of expert rules with unprecisely formulated weights understood as conditional probabilities. Adopting the probabilistic model the combined and unified recognition algorithms are derived. In the first approach algorithm is based simply on the both set of data, in the second however, one set of data is transformed into the second one. Proposed algorithms were applied practically to the diagnosis of acute renal failure in children. Obtained results have proved its effectiveness in the computer medical decisionmaking.

A branch-and-bound algorithm for optimization of multiperspective classifier

The present paper is devoted to the multiperspective recognition, in which the pattern to be recognized undergoes several classification tasks. Each task denotes here recognition from a different point of view and with respect to a different set of classes. In the decomposed dependent approach, when the multiperspective recognition is not a single activity but it states the multistep decision process, important role plays the order of recognition tasks determining the successive steps of the entire multiperspective recognition. In this paper the algorithm for optimal (with respect to the risk function) ordering of recognition tasks is presented. The proposed algorithm using controlled enumeration through a branch-and-bound search procedure selects the best order without exhaustive search. Furthermore, results of computer experiments are given and a simple illustrative example is considered.

Rule-based medical diagnosis with learning: Application to the diagnosis of acute renal failure in children

This paper presents the original concept of recognition which simultaneously uses the information contained in the learning sequence and expert rules. This method has been implemented in the computer system of the general purpose. Results of its application to the diagnosis of acute renal failure in children are described.

The bayes-optimal feature extraction procedure for pattern recognition using genetic algorithm

The paper deals with the extraction of features for statistical pattern recognition. Bayes probability of correct classification is adopted as the extraction criterion. The problem with complete probabilistic information is discussed and Bayes-optimal feature extraction procedure is presented in detail. The case of recognition with learning is also considered. As method of solution of optimal feature extraction a genetic algorithm is proposed. A numerical example demonstrating capability of proposed approach to solve feature extraction problem is presented.

Multiperspective recognition applied to the computer-aided medical diagnosis - a comparative study of methods

Deals with the multiperspective recognition technique applied to computer-aided decisions in medicine. For three different concepts of multiperspective classification, i.e. direct, decomposed independent and decomposed dependent approach, several decision algorithms are presented. They are: probabilistic (empirical Bayes) algorithm, nearest neighbour algorithm, fuzzy method and artificial neural network of the back propagation and counter propagation types. Proposed methods and algorithms have been applied to the computer-aided diagnosis of chronic renal failure and decisions in non-Hodgkin lymphoma. Results of experimental investigations on the real data and outcomes of the comparative analysis of the algorithms discussed are presented.

Multiperspective recognition: A tool for the computer-aided medical decision problems

The paper presents algorithms and the results of empirical investigations of the multiperspective recognition for the different concept of classifier action. Example of its application to the computer-aided diagnostic and therapeutic decisions in non-Hodgkin lymphoma is described.

Hilbert–Huang Transform Applied to the Recognition of Multimodal Biosignals in the Control of Bioprosthetic Hand

This paper deals with the problem of bioprosthetic hand control via recognition of user intent on the basis of electromyography (EMG) and mechanomyography (MMG) signals acquired from the surface of a forearm. As a method of signal parameterization the Hilbert–Huang (HH) transform is applied which is an effective tool for reduction of feature space dimension. The performance of proposed recognition method based on HH transform of EMG and MMG signals was experimentally compared against an autoregressive model of dimensionality reduction using real data concerning the recognition of five types of grasping movements. The experimental results show that the HH transform approach with root mean square of amplitude feature outperforms an autoregressive method.

The project of the telemedicine system for a family doctors' practices

The paper deals with a concept of a telemedicine system for the family doctor practices (FDP). The project offers the potential to improve access to high-quality primary health care, education of family doctors and patients. This is a project which is realised in collaboration of two scientific partners: Department of Medical Informatics from Wroclaw University of Technology Department of Family Medicine from Wroclaw Medical University. For a start the telemedicine system will be prepared for the Lower-Silesian territory in Poland.

Multitask Pattern Recognition Algorithm for the Medical Decision Support System

The paper presents algorithms of the multitask recognition for the decomposed dependent approach. First one, with full probabilistic information and second one, algorithms with learning sequence. We have focused our attention on the multitask recognition technique and its application to the computer aided diagnostic and therapeutic decision in non-Hodgkin lymphoma disease. Adequate computer system was projected. This system has been practically implemented in Department of Hematology if Wroclaw Medical Academy in Poland.