Programme of colloquia with abstracts for the Spring 2010 term

23 February 2010

Doc. RNDr. Václav Matyáš, M.Sc., Ph.D., FIMU, Brno

Identity (in computer and communication systems)

Abstract: The issue of identity verification (as one of the two main aspects of authentication) in computer and communication systems has been studied for decades, and our failures in doing this properly are a good illustration of the problem’s complexity. We have been adding something in the order of 10^8 ‘classical’ new computers and 10^10 mobile phones, car chips, smart cards and ‘smart dust’ every year. Consequently, various other problems relating to identity and identification arise. In this talk, we will examine some of these interesting problems concerning identity and authentication in computer and communication systems, and we will discuss identity-related solutions that are currently in use or being considered for the future.

Prof. Masakazu Suzuki, Kyushu University

Adaptive method for the digitisation of mathematical documents

Abstract: In our laboratory at Kyushu University, Japan, we have been developing Math OCR software InftyReader for more than ten years. It recognises scanned images or PDFs of mathematical documents, including various formulae from pure and applied mathematics, and outputs the results in various ‘editable’ document formats such as XML, LaTeX source files, MathML and Word documents, etc. It automatically segments the page images into figures, tables and text areas, and then recognises ordinary text and mathematical formulae within the text areas using different OCR engines.

In many cases, OCR is used to digitise documents comprising a large number of pages – ranging from hundreds to more than ten thousand pages, depending on the user’s purpose. During the talk, accompanied by a demonstration of our newly developed system, we will present our approach to improving recognition rates efficiently in such large-scale digitisation of mathematical documents, based on an adaptive method utilising character features and other statistical information extracted from the target documents.

2 March 2010

Academic sculptor Štěpán Pala, Bratislava

Reflections on Geometry

Abstract: In this lecture on the relationship between geometry and the visual arts, attention will be paid to:

1. the syntax of basic geometric solids, spatial growths, and open and compact geometry,
2. the movement and transformation of basic geometric structures,
3. the construction of spheres,
4. a theoretical conclusion and future directions in this field.

The lecture will be followed at 4.00 pm by the opening of an exhibition of selected works by the lecturer at the Bezejmenná Gallery. Everyone is welcome.

9 March 2010

Doc. RNDr. Rastislav Královič, Ph.D., Faculty of Mathematics, Physics and Informatics, Comenius University, Bratislava

Online computations: When is a crystal ball worthwhile?

Abstract: We use the term ‘online’ to describe calculations in which part of the output must be determined before the entire input has been read; the relevant algorithms therefore operate without knowledge of the complete input information. A typical example is memory paging : when a page is requested that is not currently in physical memory (a so-called ‘page fault’), it is necessary to evict an existing page from memory and replace it with the requested one, whilst the page to be evicted must be selected without knowledge of subsequent requests. The standard measure for evaluating the quality of online algorithms is the so-called competitive ratio, i.e. the ratio of the cost of the online solution to the worst-case optimal solution. The competitive ratio thus indicates the cost (in terms of the solution’s cost) of not knowing the complete input. In the first part of the lecture, we will briefly mention the basic definitions and several typical results from the field of online computations.

The second part of the lecture will be devoted to recent results that investigate the amount of information in the unknown part of the input that is relevant to a given problem. These are models in which, in addition to the input read so far, input, a limited number of bits of auxiliary information (advice) about the entire input, whilst considering the best-case scenario for the auxiliary information. We are interested in how the size of the auxiliary information affects the quality of the resulting solution.

16 March 2010

Prof. Hermann Maurer, Graz University of Technology,
Chairman of the Informatics Section of Academia Europaea

Theory is important but can be dangerous

Abstract: In this talk, I explain that many breakthroughs in applied science have been driven by new theoretical discoveries in mathematics and computer science; hence, fundamental research is of crucial importance for technical progress. I will present some typical examples. The downside is that some theoretical results have also discouraged researchers from seeking solutions where, despite negative theoretical findings, very useful solutions did in fact exist. Finally, a dangerous effect is discussed: learning is not always beneficial; indeed, it can conceal important knowledge!

23 March 2010

Doc. Dr. Ing. Pavel Zemčík, FIT, Brno University of Technology

New Methods in Hologram Synthesis

Abstract: This lecture will provide an overview of the hologram synthesis process. It will identify those parts of the process that are the subject of research in general and at FIT, and for selected parts it will outline the procedure, prospects for solutions and future outlook, specifically concerning scene representation using a ‘point cloud’, the calculation of the optical field of a point light source and local illumination models. An assessment of the computational complexity of the process and its feasibility will also be carried out.

30 March 2010

Prof. Rudolf Freund, TU Wien

Current trends in molecular computing

Abstract: In this talk, some of the latest developments in the field of membrane computing will be presented, including several open problems that have emerged only recently. Furthermore, some connections between the fields of membrane computing and DNA computing will be demonstrated.

6 April 2010

Assoc. Prof. PhDr. Karel Pala, CSc., Faculty of Informatics, Masaryk University, Brno

Can a computer correct our mistakes?

Abstract: What kinds of mistakes do people make in texts, and how can they be corrected by a computer? Classification of errors in texts and the creation of the ‘Errors’ text corpus based on written assignments by computer science students in the VB000 course. The corpus includes annotations of spelling, typographical, grammatical (syntactic), semantic, lexical and stylistic errors. The OOCorr tool was developed for annotation and is used by students on the VB000 course. Computer tools for correcting errors – spelling and grammar checkers, the current state of the art, their effectiveness and limitations.

13 April 2010

Prof. RNDr. Jaroslav Koča, DrSc., Faculty of Science, Masaryk University, Brno,
National Centre for Biomolecular Research

Chemistry, computational chemistry, info-chemistry – who comes out on top?

Abstract: Computational chemistry is a discipline that has existed for several decades. One of its key objectives is to assist in the interpretation of various categories of experimental data. However, the main aim of computational chemistry is to predict various aspects of a chemical system without existing experimental data, or in cases where it is not possible to carry out an experiment. The speaker will seek to answer the question posed in the lecture title by drawing on examples of research projects being carried out at the National Centre for Biomolecular Research.

20 April 2010

RNDr. Petr Holub, Ph.D., Faculty of Computer Science, Masaryk University, Brno

Network-Based Collaborative Environments

Abstract: Due to globalisation in both business and academia, distributed collaborative environments have become commonplace for the efficient management of distributed teams working together. From a research perspective, distributed collaborative environments constitute a highly multidisciplinary applied field that intersects with psychology, user interfaces and human-computer interaction in general, multimedia processing, and networking. This talk will review advances in collaborative environments made possible by changes in networking and distributed infrastructures. Research carried out at the Laboratory of Advanced Networking Technologies will also be described in this context. Future research directions based on emerging technologies, such as on-demand network circuits and massively parallel programmable hardware, together with novel user interfaces, will be discussed.

27 April 2010

Prof. RNDr. Ivana Černá, CSc., Faculty of Informatics, Masaryk University, Brno

Formal Verification of Component System Interactions – Motivation, Methods & Challenges

Abstract: Component-based software engineering addresses the issue of building large systems by combining existing and new components. Establishing the correctness of such systems is a challenging task. Among the formal analyses required to ensure global correctness, component interaction stands out as one of the key problems. Over the years, various mechanisms have been proposed to tackle this problem. In this talk, we focus on automata-based component and interaction specifications. This approach provides an elegant framework for capturing and automatically verifying the interactive behaviour of component-based systems. We address all aspects of the verification process, from system modelling and property specification, through verification methods, to the interpretation of the resulting observations.

4 May 2010

Doc. RNDr. Petr Sojka, Ph.D., Faculty of Informatics, Masaryk University, Brno

Towards a Digital Mathematics Library

Abstract: I will begin with the success story of the DML-CZ http://dml.cz library, and with the vision and objectives of the European Digital Mathematics Library project. I will describe the challenges of mathematical OCR, indexing and search, digital library visualisation, and other problems in document engineering and how we have tackled them in the DML-CZ and in EuDML.

11 May 2010

Doc. RNDr. Michal Kozubek, Ph.D., CABO, Faculty of Informatics, Masaryk University, Brno

From autofocusing to automatic 3D information content analysis

Abstract: The autofocus feature has become an indispensable part of both consumer and professional cameras. Scientific cameras typically rely on software autofocusing performed online by a host computer. The autofocusing procedure may be tailored to specific applications and implemented in different ways by different users. Nevertheless, the common strategy of all software autofocus routines is to acquire a sequence of low-resolution 2D images, such that each image is taken at a different focal plane position. For each image in this sequence, the degree of focus is computed and the position with the maximum value is determined.

In practice, however, there may be more than one significant focal plane, and less prominent objects are often overlooked during acquisition. We have optimised the autofocusing procedure and developed an algorithm that utilises the low-resolution image sequence acquired during the autofocusing procedure to perform automatic 3D information content analysis of the entire image volume and to identify all important spatial regions of interest that warrant subsequent high-resolution 2D or 3D acquisition. The approach is demonstrated using examples from optical microscopy.

18 May 2010

Mgr. Mário Ziman, Ph.D., Slovak Academy of Sciences, Bratislava; Faculty of Informatics, Masaryk University, Brno

Complexity of Physics, Physics of Complexity

Abstract: We will observe that quantum systems are, in general, difficult to simulate on classical computers. This obstacle has led to the idea of quantum-based computers. Using simple examples, we will illustrate the basic principles determining their computational power. Furthermore, we will focus on particular problems in physics that could prove difficult even for quantum computers. In fact, the potential simplicity of these problems would have interesting implications for complexity theory.

12 October 2010

Doc. Ing. Petr Tůma, Dr., Faculty of Mathematics and Physics, Charles University, Prague

Software Performance: Can It Be Modelled?

Abstract: Alongside its functionality, software performance is also an important property. In addition to methods for developing software with specific functions (requirements modelling, functional testing, formal verification, etc.), similar methods are therefore being sought for developing software with specific performance characteristics. These include software performance modelling, which can be effectively combined with other modelling approaches (MDD): This seminar therefore presents the current state of software performance modelling and methods for validating the accuracy of such models, whilst also focusing on the properties of modern architectures (multicore, virtualisation) that are currently difficult to model. The seminar partly presents the results of the EU’s 7th Framework Programme project Q-ImPrESS.