Dominik Slezak

Rough Computing: Theories, Technologies and Applications

Rough set theory is a new soft computing tool which deals with vagueness and uncertainty. It has attracted the attention of researchers and practitioners worldwide, and has been successfully applied to many fields such as knowledge discovery, decision support, pattern recognition, and machine learning. Rough Computing: Theories, Technologies and Applications offers the most comprehensive coverage of key rough computing research, surveying a full range of topics from granular computing to pansystems theory. With its unique coverage of the defining issues of the field, this commanding research collection provides libraries with a single, authoritative reference to this highly advanced technological topic.

Rough Set Approach to the Survival Analysis

Application of rough set based tools to the post-surgery survival analysis is discussed. Decision problem is defined over data related to the head and neck cancer cases, for two types of medical surgeries. The task is to express the differences between expected results of these surgeries and to search for rules discerning different survival tendencies. The rough set framework is combined with the Kaplan-Meier product estimation and the Coxs proportional hazard modeling.

Searching for the Complex Decision Reducts: The Case Study of the Survival Analysis

Generalization of the fundamental rough set discernibility tools aiming at searching for relevant patterns for complex decisions is discussed. As an example of application, there is considered the post-surgery survival analysis problem for the head and neck cancer cases. The goal is to express dissimilarity between different survival tendencies by means of clinical information. It requires handling decision values in form of plots representing the Kaplan-Meier product estimates for the groups of patients.

Fuzzy Reals with Algebraic Operations: Algorithmic Approach

Fuzzy counterpart of real numbers, called fuzzy numbers (reals), are investigated. Their membership functions satisfy conditions similar to quasi-convexity. In order to operate on them in a similar way to real numbers revised algebraic operations are introduced. At first four operations between fuzzy and real numbers are in use in a form suitable for their algorithmisations. Two operations: addition and subtraction between fuzzy numbers are proposed to omit some drawbacks of the corresponding operations originally defined by L. A. Zadeh with the help of his extension principle.

Approximation Degrees in Decision Reduct-Based MRI Segmentation

Segmentation of magnetic resonance images (MRI) is a process of assigning the tissue class labels to voxels. One of the main sources of segmentation error is the partial volume effect (PVE), which occurs most often with low resolution images. Indeed, for large voxels, the probability of a voxel containing multiple tissue classes increases. We have utilized a classification approach based on the attribute reduction, derived from the data mining paradigm of the theory of rough sets. An approximate reduct is an irreducible subset of features, which enables to classify decision concepts with a satisfactory degree of accuracy in the training data. The ensembles of the best found reducts trained for appropriate approximation degrees are applied to segmentation of previously unseen (parts of) images. One of the challenges is to adjust the approximation level during the training phase to obtain the best classification results for new cases. In this paper, it is proved experimentally that the choice of approximation level, consequently related to generality of classification rules induced by reducts, should correspond to expected quality of images. We show that when dealing with noisy images, or images with high PVE level, better results are obtained with higher degrees of approximation.

KDD-Based Approach to Musical Instrument Sound Recognition

Automatic content extraction from multimedia files is a hot topic nowadays. Moving Picture Experts Group develops MPEG-7 standard, which aims to define a unified interface for multimedia content description, including audio data. Audio description in MPEG-7 comprises features that can be useful for any content-based search of sound files. In this paper, we investigate how to optimize sound representation in terms of musical instrument recognition purposes. We propose to trace trends in evolution of values of MPEG-7 descriptors in time, as well as their combinations. Described process is a typical example of KDD application, consisting of data preparation, feature extraction and decision model construction. Discussion of efficiency of applied classifiers illustrates capabilities of further progress in optimization of sound representation. We believe that further research in this area would provide background for automatic multimedia content description.

An Automated Multi-spectral MRI Segmentation Algorithm Using Approximate Reducts

We introduce an automated multi-spectral MRI segmentation technique based on approximate reducts derived from the data mining paradigm of the theory of rough sets. We utilized the T1, T2 and PD MRI images from the Simulated Brain Database as a ”gold standard” to train and test our segmentation algorithm. The results suggest that approximate reducts, used alone or in combination with other classification methods, may provide a novel and efficient approach to the segmentation of volumetric MRI data sets.

Application of Normalized Decision Measures to the New Case Classification

The optimization of rough set based classification models with respect to parameterized balance between a models complexity and confidence is discussed. For this purpose, the notion of a parameterized approximate inconsistent decision reduct is used. Experimental extraction of considered models from real life data is described.

Constructing Extensions of Bayesian Classifiers with Use of Normalizing Neural Networks

We introduce a new neural network model that generalizes the principles of the Naive Bayes classification method. It is trained with use of backpropagation-like algorithm, in purpose of obtaining optimal combination of several classifiers. Experimental results are presented.

Rough Set Theory in Conflict Analysis

The importance of multi-agents systems, models of agents’ interaction is increasing nowadays as distributed systems of computers started to play a significant role in society. An interaction occurs when two or more agents, which have to act in order to attain their objectives, are brought into a dynamic relationship. This relationship is the consequence of the limited resources which are available to them in a situation. If the number of resources is insufficient to attain agents’ goals it often comes into the conflicts. This can happen in almost all industrial activities requiring distributed approach, such as network control, the design and manufacture of industrial products or the distributed regulation of autonomous robots. However, distributed systems is only one from many different areas where a conflict can arise and where it is worth to apply computer aided conflict analysis. Just to mention some human activities like business, government, political or military operations, labour-management negotiations etc. etc.