Program seminářů pro rok 2005/2006

Podzim 2005

22. 9. 2005: Úvodní seminář podzimního semestru; Program:
Informace o koncepci semináře v podzimním semestru.
Domluva programu semináře.
Diskuse.
29. 9. 2005: P. Cenek; A Framework for Rapid Multimodal Application Design; Abstract:
In the talk, my collaborative work with Swiss Federal Polytechnical Institute in Lausanne (EPFL) will be described. The aim of the work is to extend the EPFL unimodal dialogue platform with multimodal capabilities. The EPFL dialogue platform will be introduced together with EPFL Rapid Dialogue Prototyping Methodology (RDPM). Then, our approach to its extension will be explained. We formulated a set of design principles that should lead to an efficient multimodal interface for a given application. The principles have been tested on a prototype application in a series of Wizard of Oz experiments. The results will be presented together with a short practical demonstration of the application and the Wizard of Oz interface.
6. 10. 2005: L. Bártek; Optimization of Generated Dialogues; Abstract:
When dialogues are generated from the graphical user interface, the optimality of the resulting dialogue might be unsatisfactory. This talk will propose some methods allowing us to evaluate the dialogue optimality and to optimize the resulting dialogue according to the proposed criterion.
13. 10. 2005: I. Peterlík; Finite Element Method and Soft Tissue Modelling; Abstract:
Virtually every phenomenon in nature can be described with the aid of the laws of physics, in terms of algebraic, differential or integral equations. Since it is often difficult or even impossible to find the exact solution of such equations, numerical methods play an important role in simulations and modelling. The talk consists of two parts. First, we present an introduction to the Finite Element Method, which provides a powerful tool for solving complex physical problems. In the second part we focus on medical simulations, namely soft tissue modelling. We look at design of surgical simulator with haptic interaction and describe the application of finite element method in this area.
20. 10. 2005: M. Batko; Scalability in Distributed Similarity Searching; Abstract:
As the growth of digital data accelerates in variety and extent, huge data repositories are becoming available on contemporary computer networks. For users to be able to access selected data objects, the objects need to be structured and manipulated efficiently but also effectively. In contrast to traditional databases made up of simple attribute data, contemporary data is bulkier and more complex in nature. For such domains, an exact match has little meaning and proximity concepts (similarity, dissimilarity) are typically much more fruitful for searching. Another problem, we should address nowadays, is the scalability of structures supporting the efficient data retrieval. Creating a search index structure which scales to very large dimensions, presents many challenges and the task is becoming increasingly difficult as the amount of data grows. Probably the most successful search engine which scales even to the Web dimension, is the Google, but it can only manage text-like data. We show that recent trends in network architectures, such as the peer-to-peer communication paradigm, can also be exploited to develop real scalable and distributed similarity search structures for arbitrary metric distance functions. The presentation will provide a brief insight into four examples of such structures.
27. 10. 2005: H. Mlnařík; Towards a Categorical Semantics of a High-level Quantum Programming Language; Abstract:
Quantum mechanics gets more and more important in computer science. Even though it is nowadays mainly theoretical field it will probably become a necessity for practice in one or two decades. We present a proposal of a category which can serve for definition of a semantics of a programming language that combines both quantum and classical computations.
3. 11. 2005: M. Kuba; Semantic Grid; Abstract:
Semantic Grid is an extension of the current Grid, where information and services are given a well-defined meaning. As an introduction to this area, an overview of current Grid technologies will be presented, together with the concept of ontologies and semantic descriptions. Then our work on using semantic grid infrastructure for solving problems in biomedicine will be presented.
10. 11. 2005: O. Krajíček; SEAGRIN - Adaptine Grid; Abstract:
TBA
24. 11. 2005: O. Výborný; Privacy in Data Mining; Abstract:
In this talk, an overview of problems related to privacy in data mining will be presented together with different approaches for solving these problems. We will focus on different points of view on privacy and show several methods used for preserving the privacy. We will also show some useful utilizations of data mining for IT security.
1. 12. 2005: J. Hubený; 3D Reconstruction of Interphase Chromosome Territories; Abstract:
In this talk, we describe a robust and mathematically well founded segmentation method suitable for analysis of image data acquired using confocal microscope. We will focus on the specific problem of reconstruction and segmentation of interphase chromosome territories. The proposed method uses PDE-based nonlinear filters (namely the Balanced Forward Backward filter) for image preprocessing and PDE-based implicit active contour models (Chan-Vese segmentation model) for object reconstruction. Both these techniques are not widely used in image analysis in confocal microscopy because they are quite complex and their computation can be very slow when processing large three dimensional data. Nevertheless, we will show that one can overcome slow computation speed, which is the main disadvantage of PDE-based image processing techniques, by using fast and sophisticated numerical approximations and by processing the input image per partes. Last but not least we will discuss the performance and properties of proposed reconstruction method and our plans to the future. We will finish the talk with the practical demonstration of the software, which implements the proposed method.
8. 12. 2005: P. Adámek; Indexing XML Documents that Change; Abstract:
Many indexing methods have been developed to allow efficient searching collections of XML documents. However, many of them are static and do not efficiently support updates of the indexed documents. This talk discusses this problem and offers some solutions. Further, we present an approach which is based on XML Tree Signatures.
15. 12. 2005: T. Rebok; QoS-enabled Distributed Multistream Active Router; Abstract:
Active Networks (Active Routers) are suitable approach for giving higher flexibility to current computer networks enabling user to upload their own code for data processing inside the computer network. To introduce this approach into high-speed computer networks a distributed environment and Quality of Service (QoS) assurance are necessary. In the first part of this talk some requests on such router and its possible practical application will be shown and in the second part we will briefly discuss problems of deploying Active Networks on current computer networks and their assumed future.
22. 12. 2005: Poster Session; Abstract:
Prezentovány budou postery vytvořené v podzimním semestru.
2. 3. 2006: P. Moravec; Distributed Decompositions of Graphs into Strongly Connected Components; Abstract:
Decomposition of graphs into strongly connected components (SCCs) is a classic graph-theoretic problem with applications in many areas involving formal verification. Since distributed verification methods have became very popular, there is the natural need to propose distributed algorithms for the SCCs decomposition. In this talk I present novel algorithms solving that problem and discuss its properties.
9. 3. 2006: L. Bártek; Generating Web-based Presentations by Means of Dialogue; Abstract:
The talk deals with problems that may occur during the process of generating web pages by means of dialogue. Some topics related to the generation of web-based presentation will be discussed as well.
16. 3. 2006: P. Medek; Analysis of Protein Structure; Abstract:
Identification of export tunnels leading from the inside of a protein molecule to its surface is believed to bring a significant improvement in the area of drug design. We will describe how this problem has been solved up to the present and propose a new solution based on Voronoi diagrams. Mainly we will focus on a temporary hybrid algorithm combining classical Voronoi diagrams computation with Euclidean Voronoi diagrams. At the end of this talk we will discuss the way in which this method could be extended to compute an exact solution.
23. 3. 2006: P. Očenášek, FIT VUT; Evolutionary Approach in the Security Protocol Design; Abstract:
This presentation is devoted to an evolutionary method that serves for designing security protocols. The principles of security protocols are outlined, followed by the reasoning about modal logic that is used to encode the belief and knowledge of communicating parties. The second part introduces the evolutionary optimization framework and proposes the techniques that can be used to automatically evolve security protocols.
30. 3. 2006: J. Špojcar; Relations between Anonymous Channels; Abstract:
We take a closer look at anonymous channel - the most simple application of anonymity in cryptography. An anonymous channel is a primitive for communication between parties. Every anonymous channel achives some kind of anonymity e.g. anonymous sender, relationship anonymity, unlinkability message - receiver etc. We present many types of anonymous channels as well as a structure (a lattice) that describes some relations between them - simulation of one channel by another one. We call this relation reduction.
6. 4. 2006: L. Boháč; Superstrong Correlations in Quantum Cryptography; Abstract:
Classical as well as quantum theories put limits on non-local correlations that can be generated between two separated sites when no communication is allowed. Concept of superstrong correlations, maximal correlations allowed by special relativity, as well as their possible impact on quantum cryptography will be presented.
13. 4. 2006: M. Bidlo; Evolutionary design using development; Abstract:
The presentation is devoted to two different models of development for the evolutionary design of combinational logic circuits. The first of them, an instruction-based approach, is able to construct arbitrarily large circuits of several classes. Various examples of evolved circuits will be shown and discussed (e.g. sorting networks, polymorphic circuits). The second method utilises cellular automaton for the design of combinational circuits of a given number of inputs. Finally, a plan of possible future research activities will be given.
20. 4. 2006: J. Denemark; Virtual Grid Execution Environmant; Abstract:
The talk is focused on virtualization and its usage in grid environment. We will show how it can be beneficial for grids and how it can help with satisfying heterogeneous requirements of users. An overview of some existing technologies will be presented and an architecture of a virtual execution environment we designed will be described.
27. 4. 2006: P. Vařeková; Specification of Software Components with Component-Interaction Automata; Abstract:
The talk presents a verification-oriented automata-based language, Component-Interaction automata, which is designed for specification of component interactions in hierarchical component-based software systems. First, we mention software components and their interaction properties. Then we will focus on a definition of Component-Interaction automaton. The last part of the talk discusses modeling composition of components in specification language Component-Interaction automata.
4. 5. 2006: A. Rogalewicz; Verification of Programs With Dynamic Data Structures by ARTMC; Abstract:
For verification of programs working with dynamic data structures (e.g. SLL, DLL, Trees, ...), it was proposed an encoding of a program behavior into tree transducers. Configurations of such programs are encoded as trees, and sets of such configurations as tree automata. This encoding allows us to use "abstract regular tree model checking method". The complete method was prototypelly implemented over tree libraries from the MONA project. This method was successfully used on several examples.
11. 5. 2006: P. Adámek; TBA; Abstract:
TBA
18. 5. 2006: Poster Session; Abstract:
Prezentovány budou postery vytvořené v jarním semestru.