Catálogo de publicaciones - libros

We discuses usage of elements of. Net platform — Web Service and XML to create distributed internet applications as data processing system. The main aim is to create retrieval system of hidden relations between data. The application uses elements of rough sets theory in order to get data, which could be used for further exploration. Data are sent to particular Web Services as XML format and could represent various domains e.g. medicine, pharmacy. The implementation intelligent techniques of data processing based on Web Services and XML standard, illustrates new possibilities for data processing in distributed environment. It gives scaling and possibility to adapt it in many specialized solutions.

There has been interest recently in skyline queries, also called Pareto queries, on relational databases. Relational query languages do not support search for “best” tuples, beyond the order by statement. The proposed skyline operator allows one to query for best tuples with respect to any number of attributes as preferences. In this work, we explore what the skyline means, and why skyline queries are useful, particularly for expressing preference. We describe the theoretical aspects and possible optimizations of an efficiant algorithm for computing skyline queries presented in [6].

Grouping data into meaningful clusters belongs to important tasks in the area of artificial intelligence and data mining. DBSCAN is recognized as a high quality scalable algorithm for clustering data. It enables determination of clusters of any shape and identification of noise data. In this paper, we propose a method improving the performance of DBSCAN. The usefulness of the method is verified experimentally both for indexed and non-indexed data.

On October 10, 2001, the US President identified the most wanted persons sought by the United States of America. Agencies such as FBI, CIA and Homeland Security spread images of the most wanted persons across the United States. Even though US citizens saw their images on television, the Internet and posters, computers had, and still have for that matter, no ability at all to identify these persons. To date FBI, CIA and Homeland Security depend entirely on human beings, not computers, to identify persons at borders and international airports. In other words, facial recognition remains an incompetent technology.

Accordingly, authors first succinctly show the weaknesses of the current facial recognition methodologies, namely Eigenface Technology, Local Feature Analysis (from the classical 7 point to the 32–50 blocks approach), the Scale-Space Approach, Morphological Operations and industrial or patented methodologies such as ILEFIS™, Viisage™, Visionics™ and Cognitec’s FaceVACS-Logon™, Identix™ and Neven Vision™. Secondly, they introduce a completely new, simple and robust methodology called .

The first part of the Innertron methodology for facial recognition, presented in [1], removes those faces that certainly do not fall within the constraints of the nose domain with an upper limit at (2.0, −1.5) to (−2.0, −1.5) and a lower domains at (3.0, −7.5) to (−3.0, −7.5). The 300 square pixels will contain the nose of the subject sought. The second part of the Innertron methodology, presented in this paper, in essence knows that the subject sought is somewhere and now starts looking at features such as the eyebrows, angles of the eyes, thickness of the nose, shape of the nostril. Seven regions are distinguished that take into account the ability to lift eyebrows and most importantly the ability to cover portions of the face and still garner a in the database.

Approximating a collection of patterns is a new and active area of research in data mining. The main motivation lies in two observations : the number of mined patterns is often too large to be useful for any end-users and user-defined input parameters of many data mining algorithms are most of the time almost arbitrary defined (e.g. the frequency threshold).

In this setting, we apply the results given in the seminal paper [11] for frequent sets to the problem of approximating a set of approximate inclusion dependencies with . Using the fact that inclusion dependencies are “representable as sets”, we point out how approximation schemes defined in [11] for frequent patterns also apply in our context. An heuristic solution is also proposed for this particular problem. Even if the quality of this approximation with respect to the best solution cannot be precisely defined, an interaction property between IND and FD may be used to justify this heuristic.

Some interesting perspectives of this work are pointed out from results obtained so far.

Recent results on the evolution of the genetic code are informally put in the context of Von Neumann’s theory of self-reproducing automata and the classical treatment of error-correcting codes in information theory. I discuss the sufficiency of genetic descriptions for self-reproduction. This implies a duality, a division of the information for maintenance and reproduction of organisms into two components, which is fundamental to life as we know it. The two components are programmatic and operational in nature, and they require the transmission and transduction of information to cooperate. In this context, I review recent results on the coevolution of genes and genetic codes. These results show how desirable informatic properties can evolve in systems of tapes and tape-players that are mutually co-dependent for their reproductive success.

We propose a novel approach to discover useful patterns from ill-defined decision tables with a real value decision and nominal conditional attributes. The proposed solution is based on a two-layered learning algorithm. In the first layer the preference relation between objects is approximated from the data. In the second layer the approximated preference relation is used to create three applications: (1) to learn a ranking order on a collection of combinations, (2) to predict the real decision value, (3) to optimize the process of searching for the combination with maximal decision.

Amino acid similarity matrices are used for protein sequence comparison. A new method for studying connections between amino acid properties and similarity matrices is proposed. A representation of the amino acid similarity matrices, by means of the equivalence classes between amino acids, is proposed. It is shown that mappings between these equivalence classes and contiguous subsets of the amino acid property space are possible. These mappings are consistent between equivalence classes. There is a possibility of practical applications of this method in sequence similarity searches.

The paper presents a computer experiment inspired by the immune metaphor and based on the work of Farmer, Packard, and Perelson [FPP86]. We develop a model influenced by the way the immune system works that is well-suited to address a particular class of NP-hard problems. We discuss the results obtained when applying the model to an artificial vision problem denoted , where artificial agents successfully accomplish a surveillance assignment to protect the pieces of an art exhibition from bad behaved visitors.