Catálogo de publicaciones - libros

Emotions can be expressed in various ways. Music is one of possible media to express emotions. However, perception of music depends on many aspects and is very subjective. This paper focuses on collecting and labelling data for further experiments on discovering emotions in music audio files. The database of more than 300 songs was created and the data were labelled with adjectives. The whole collection represents 13 more detailed or 6 more general classes, covering diverse moods, feelings and emotions expressed in the gathered music pieces.

The problem of testing statistical hypotheses of independence of two multiresponse variables is considered. This is a specific inferential environment to analyze certain patterns particularly for the questionnaire data. Data analyst normally looks for certain combination of responses being more frequently chosen than the other ones. As a result of experimental study we formulate some practical advices and suggest areas of further research.

Artificial immune systems turned out to be interesting technique introduced into the area of . In the paper the idea of an immunological selection mechanism in the agent-based evolutionary computation is presented. General considerations are illustrated by the particular system dedicated to function optimization. Selected experimental results conclude the work.

Nowadays detecting destructive attacks and dangerous activities is crucial problem in many real world security systems. A security system should enable to distinguish some actors which effects of behavior are perhaps unfavorable for a considered area. The considered areas are real world systems e.g. airports, shops or city centers. Project of real world security system should assume the changing and unpredictable type of dangerous activities in real world systems. Security system has to detect and react to new kind of dangers that have never been encountered before. In this article there are presented methods derived from some ethically-social and immunological mechanisms that should enable automated intrusion detection.

Today’s text searching mechanisms are based on keyword search. However a large number of results makes the searches less effective. This paper presents results when search of relevant documents is obtained based on currently browsed document instead of keywords. Five keywords based search methods were tested. Comparative study was done on following methods: LSA, PLSA, WebSOM, PHITS, Query Dependant PageRank.

Much work in AI is devoted to the problem of how computers can acquire beliefs about the world through perception, but little effort is devoted to the problem of how computers can acquire beliefs through testimony. This paper is part of a continuing project whose ultimate goal is that of constructing an implementable model of how agents acquire knowledge through testimony. In particular, it looks at how agents acquire information from the radio and many factors are identified that may cause an agent to override the defeasible rule to believe what he hears.

An in-house developed computer program system , capable to generate belief networks and to convert them into respective sets of belief rules, was applied in mining the melanoma database. The obtained belief rules were compared with production rules generated by system. It was found, that belief rules can be presumably treated as a generalization of standard rules.

This paper discusses a new design for the matchmaker used in DECAF

For machine learning of an input-output function from examples, we show it is possible to define an a priori probability density function on the hypothesis space to represent knowledge of the probability distribution of , even when the hypothesis space is large (i.e., nonparametric). This allows extension of maximum a posteriori (MAP) estimation methods nonparametric function estimation. Among other things, the resulting MAPN (MAP for nonparametric machine learning) procedure easily reproduces spline and radial basis function solutions of learning problems.

Classic kernel methods (SVM, LS-SVM) use some arbitrarily chosen loss functions. These functions equally penalize errors on all training samples. In problem of time series prediction better results can be achieved when the relative importance of the samples is expressed in the loss function. In this paper an autocovariance based weighting strategy for chaotic time series prediction is presented. Proposed method can be considered a way to improve the performance of kernel algorithms by incorporating some additional knowledge and information on the analyzed learning problem.