Once a week during the academic year, a guest speaker (from abroad or from the Czech Republic) gives a lecture on their research. The colloquium, which takes place at the Faculty of Informatics, is open to the academic community. You can find the lecture dates in the programme. Tuesday 14.00–15.00, D2, FI MU, Botanická 68a

Colloquium programme with abstracts for the Spring 2015 semester

17 February 2015

Doc. Mgr. Radek Pelánek, Ph.D., FI MU

Discussion on the revision of degree programmes

Abstract: Dear colleagues, Over the past few months, discussions have taken place (within the working group, management and the academic board) regarding the long-term revision of study programmes (for the next accreditation period). Next month, there will be a broader presentation and discussion on this topic. Outline of the programme: - clarification of the context (the current status of accreditations, the outlook, the timetable) - presentation of a working draft arising from the discussions to date - presentation of the main options, comments and unresolved issues - discussion – open not only to comments on the draft presented, but also, for example, to ‘visionary’ proposals for areas we should aim to cover The discussion will take place at the first seminar of the term, i.e. on Tuesday 17 February. I am letting you know in advance so that you can set aside the time and so that you have time to think about your visionary proposals.

24 March 2015

Fotios Liarokapis, Ph.D., FI MU

Procedural Generation for Interactive Virtual Environments

Abstract: The creation of realistic virtual environments is a key issue in the computer animation, computer games, digital film effects and simulation industries. Whilst many companies have the necessary budget to develop these expensive, modern computer games that utilise state-of-the-art computer graphics, not all developers have the same resources. Consequently, procedural generation is an area of growing interest within the field of computer graphics and virtual environments. This presentation will provide an overview of procedural generation methods and techniques for both content and human behaviour. In terms of content, the main focus will be on generating realistic terrain environments, as well as buildings and, subsequently, cities. As far as human behaviour is concerned, the main emphasis will be on accurately modelling crowds in urban environments. Specific examples will be demonstrated in the field of virtual reconstruction of cultural heritage, as well as in interactive computer games.

Brief CV: Dr Fotis Liarokapis is an Associate Professor at the Human-Computer Interaction (HCI) Laboratory, Masaryk University. His research interests include: virtual and augmented reality; computer graphics; human-machine interaction; and serious games. He has contributed to more than 85 peer-reviewed publications and has more than 1,000 citations (h-index: 19 and i10-index: 29). He has been invited more than 80 times to serve on international conference committees and has chaired 14 sessions at 9 international conferences. He has secured around £200,000 in funding from a number of research projects and is a member of IEEE, IET, ACM and Eurographics. He is the co-founder of the IEEE VS-Games conference and has organised a number of other conferences, workshops and special issues in peer-reviewed journals. For further information, visit: https://www.fi.muni.cz/~liarokap/.

3 March 2015

Doc. RNDr. Tomáš Pitner, Ph.D., Faculty of Informatics, Masaryk University, and the CERIT Science Park team

CERIT Science Park

Abstract: CERIT Science Park is Masaryk University’s first science and technology park and one of the key components of the Faculty of Informatics’ recent development phase. The aim of the lecture is to outline the park’s role as a platform for long-term applied collaboration between the faculty’s research teams and the Institute of Computer Science and companies in the IT sector. The ideas and objectives behind its establishment and implementation will be outlined, and its current tenants and facilities will be presented. The potential for scientific and research collaboration will be illustrated using examples of selected companies and sectors.

10 March 2015

Assoc. Prof. RNDr. Tomáš Brázdil, Ph.D., Faculty of Informatics, Masaryk University

Stochastic Systems with Counters

Abstract: Stochastic systems with counters are a general formalism suitable for modelling systems with queues, resources, etc., which either operate in a random environment or utilise randomisation themselves (for example, to optimise performance). In this lecture, I will briefly introduce these systems and outline their connection to stochastic Petri nets and queueing networks. I will then focus on solving one of the fundamental problems in this area: calculating the probability of a counter reaching zero in one or more counters with a given probability. Reaching the value zero in a system with counters corresponds to emptying a place in a stochastic Petri net, or emptying a queue in a queueing network. From a practical point of view, this may, for example, involve the depletion of a given resource or, conversely, the successful completion of all tasks waiting in the queue. In this lecture, I will present fundamental theoretical results concerning the decidability and complexity of the problem of reaching zero. I will also present results concerning upper and lower bounds on the probability of reaching zero.

17 March 2015

Prof. RNDr. Jiří Wiedermann, DrSc., Institute of Computer Science, Czech Academy of Sciences

Towards Computational Principles of Creativity

Abstract: The principles of creativity are investigated within the framework of the observer-dependent definition of computation recently introduced by the author and Jan van Leeuwen. Unlike classical definitions of computation, the new definition is machine-independent and algorithm- and representation-free. It views computations as knowledge-generation processes satisfying three basic principles: repeatability, compositionality and justifiability. Each computation operates within its specific knowledge domain. Within this approach, creativity is viewed as the ability to create knowledge that solves a given novel problem. We hypothesise that the essence of creativity is a computational process that systematically generates or searches for all knowledge within a given epistemic domain until knowledge satisfying the constraints required by the problem’s solution is found. This seemingly inefficient discovery process is, in fact, a learning process that automatically extracts and modifies ‘user preferences’ on each occasion. These preferences are subsequently used to narrow the search space of the discovery process and to rank the solutions found. Cultivating the creativity processes in this way and tailoring them to specific domains, together with the exploitation of massive parallelism across large, heterogeneous epistemic bases, eventually leads to the emergence of effects familiar from human-like creative processes.

24 March 2015

Prof. Ing. Lukáš Sekanina, Ph.D., FIT VUT

Approximate computing

Abstract: The concept of approximation has been intensively and continuously studied, developed and applied not only in computer science, but also in mathematics and engineering disciplines. The ever-increasing pressure to reduce the power consumption of computing devices (particularly in the field of mobile electronics), the exponential increase in the number of transistors on a chip (and the associated increase in power density) and the ever-growing volume of data that must be processed rapidly have led to the introduction of approximations even into areas where, traditionally, only error-free solutions were acceptable. A new field of research and application is emerging, known as approximate computing, where approximations are systematically introduced at the level of programming languages, software, computing architecture and even the elementary components of processors, such as the ALU or memory storage, with the aim of reducing power consumption or speeding up computation, even at the cost of computational errors. Typical applications include multimedia, data mining, prediction and low-power components, where computational errors can occasionally be tolerated. An open question remains as to how to automatically design approximate computing systems. The lecture will introduce the field of approximate computation and provide examples of approximations (arithmetic circuits, the median) which are carried out at FIT, BUT, using genetic programming.

31 March 2015

Mgr. Pavel Rychlý, Ph.D., Faculty of Informatics, Masaryk University

Why programme in Go

Abstract: Go is a new programming language developed by Google, originally designed for large-scale programming (large projects with a large number of programmers). Go is a traditional procedural language (in the style of C) with a number of non-traditional features. For example, it is an object-oriented language, yet lacks classes, inheritance, constructors and destructors. It provides automatic memory management whilst allowing low-level programming. It is a compiled language (supporting all major platforms), and the compilation of even large-scale systems is faster than running the interpreters of some languages. Go’s approach to concurrent programming is unique and elegant, differing significantly from the conventional parallel approach using threads.

The lecture will present the language’s basic features and demonstrate its advantages for use in research. These advantages will be illustrated using the results of a reimplementation of the Manatee system carried out at the Centre for Natural Language Processing at the Faculty of Informatics, which led to clearer code and faster execution of its key parts.

7 April 2015

RNDr. Jiří Grygar, CSc., Institute of Physics, Czech Academy of Sciences

Petabyte Astronomy

Abstract: Since the invention of the telescope in the early days, the main task of astronomy has been to compile catalogues of stars and galaxies containing, initially, tens and later thousands of kilobytes of data. Photographic and spectroscopic surveys of the sky in the 20th century pushed this limit to the terabyte range. Currently, astronomical databases based on all-sky surveys are in the petabyte range. This places extraordinary demands on the efficient processing of such vast datasets. Astronomers are pioneers of distributed computing methods (e.g. SETI@home; Galaxy Zoo, etc.) and have created a global virtual observatory, containing all available data on individual astronomical objects in a compatible format. Long-baseline optical and radio interferometry places exceptionally high demands on rapid data processing, achieved through the parallel deployment of hundreds of millions of processors with speeds of the order of tens of Pflops. The instruments being designed for optical and radio astronomy over the next decade will utilise supercomputers with performance levels of >100 Pflops. At the same time, computer simulations of complex astrophysical processes are also being developed, in particular the evolution of the structure of the universe over the last 13 billion years, supernova explosions and black hole collisions, which require up to ten thousand years of CPU time.

14 April 2015

Doc. RNDr. Vlastislav Dohnal, Ph.D., Faculty of Informatics, Masaryk University

Big Data Analytics

Abstract: Many organisations today have to cope with an enormous volume of data and its variability. It is not only necessary to store the data but also to process it in order to make any use of it. In general, we refer to this as the Big Data problem. In this presentation, we will provide an overview of the principles of data analytics from a historical perspective, covering both traditional data warehouses and distributed systems capable of handling Big Data.

21 April 2015

Doc. RNDr. Eva Hladká, Ph.D., Faculty of Informatics, Masaryk University

Collaborative technologies for teaching students with hearing impairments

Abstract: Teaching students with hearing impairments is particularly challenging due to the student’s very limited ability to perceive the teacher. Meaningful interaction is only possible with the assistance of an interpreter. However, specialist interpreters, capable of translating even technical terms into sign language, are in high demand and cannot always be present in person. Video-conferencing technology can help, but a fully effective transmission requires high-quality video from both the teacher and the interpreter, so that the student is able to lip-read and follow sign language simultaneously. At the Faculty of Informatics, Masaryk University, in collaboration with CESNET, the Ultragid tool has been under development for some time for video conferencing with high-definition video. This system is a potential solution for the use of remote interpreting. However, high image quality places significant demands on network transmission in terms of parameters such as bandwidth and latency. Transmission quality can be addressed by another system developed here – Couniverse, which complements Ultragrid by monitoring connection quality and ensuring compliance with the required parameters. How can these technologies be made accessible to non-technical users within a remote interpreting system? This problem is addressed by the CoUnsil system, which I shall present at the end of the lecture.

28 April 2015

Helena Lukášová, MA, PhD, Faculty of Informatics, Masaryk University

Digital Fabrication

Abstract: The lecture ‘Digital Fabrication’ follows on loosely from the lecture on 3D printing, which took place in September last year. The aim will be to present the possibilities for using digital tools in sculpture, from the beginnings of this alliance to the present day. To provide a comprehensive understanding of the topic, examples of digital sculpture from before the advent of digital fabrication methods will also be presented. Sculptors use CNC machines and robotic machining arms to create works on a larger scale and in traditional materials. The use of these technologies significantly speeds up the creation of a work, whilst also offering new possibilities even at the stage of conceptualising the work itself. Methods of so-called digital fabrication are thus becoming both a tool and a means by which a new dimension can be brought to contemporary digital art. Until recently, the computer was a symbol of liberation from working with physical materials; however, new possibilities are disrupting this paradigm.

5 May 2015

Docent JUDr. Radim Polčák, Ph.D., Faculty of Law, Masaryk University

Cyber Security for Hedgehogs

Abstract: The structure of the Cyber Security Act contains a number of specific contradictions and problematic aspects. Nevertheless, it is undoubtedly the first comprehensive legislation of this kind to have successfully passed through the legislative process in a democratic state governed by the rule of law. The reason for the political success of the newly created legal phenomenon of cyber security lies primarily in the implicit values it embodies, which are not contradictory in nature. Whilst other components of security law are naturally contradictory in terms of values (i.e. the conflict between security and privacy, security and property, or even security and autonomy of will), the Czech legal framework is not based on the need to weigh up conflicting interests proportionately. In addition to this previously unpublished issue, the paper will also present selected problematic aspects of the new Czech legislation, with particular attention being paid primarily to the issue of compliance and the legal liability of the individual.

12 May 2015

Mgr. Marek Sýs, Ph.D., Faculty of Law, Masaryk University

Randomness testing

Abstract: Randomness plays a fundamental role in cryptography. A function of any type (hash functions, block ciphers, stream ciphers, etc.) should produce sequences with no recognisable patterns or regularities. Randomness is examined using empirical randomness tests. Tests are grouped into test suites to enable more complex analysis. In this talk, we will show you how to examine randomness using standard test suites (NIST STS, Diehard, TestU01) and how to interpret their results. The test suites implement a fixed set of tests and can detect only limited types of patterns. We also introduce a novel and more general framework for randomness analysis based on evolutionary algorithms. The framework is capable of detecting (theoretically) arbitrary types of patterns.