Programme of colloquia with abstracts for the Autumn 2012 semester

25 September 2012

RNDr. Jan Strejček, Ph.D., Faculty of Informatics, Masaryk University

Symbolic Execution and its Practical Applications

Abstract: Symbolic execution is one of the classical methods of programme analysis. This talk will present the basic principles of symbolic execution and explain its advantages and disadvantages. It will also describe some successful applications of symbolic execution in the field of software fault detection and outline current research trends in this area.

2 October 2012

Prof. RNDr. Luděk Matyska, CSc., Faculty of Informatics, Masaryk University

RNDr. Michal Procházka, Faculty of Informatics, Masaryk University

User-Centric Identity Federations

Abstract: The Internet is full of services that require authentication followed by an authorisation decision, and the number of such services is rapidly increasing. Identity federations offer a solution based on the use of the user’s home organisation, utilising its authentication mechanism (well known to end users) and the data the organisation holds about its own users. The external services can thus not only rely on authentication provided by the home organisation, but can also obtain additional data about the user, data to be used for authorisation decisions. However, whilst the use of the identity federation concept is growing successfully, its wider adoption is also revealing its shortcomings. Using a service connected to 15 national identity federations, we will discuss the limitations and present a novel solution—the Aditi system—based on a user-centred model.

9 October 2012

RNDr. Vojtěch Krmíček, Ph.D., South Moravian Innovation Centre, Brno

From idea to your own business – the StarCube programme to support master’s and PhD graduates

Abstract: Recently, we have witnessed the emergence of numerous projects and young companies in the field of new technologies and IT, known as start-ups. Various platforms and support programmes assist them in their early stages; one such programme is StarCube, run by the South Moravian Innovation Centre in Brno. The programme begins with individuals or small teams. Over the course of three months of intensive workshops, consultations with experts and mentors, and collaborative work, they have the opportunity to develop into rapidly growing companies with a well-coordinated team.

Outline:
- an introduction to JIC Brno and its activities;
- the Czech start-up scene and what a start-up is;
- the StarCube programme;
- success and failure stories, using a StarCube graduate as an example;

As well as an introduction to the StarCube programme and other activities organised by the South Moravian Innovation Centre, a successful graduate of the spring StarCube intake will also take the floor to share the experiences he gained on his journey from an idea to a functioning IT company.

16 October 2012

RNDr. Pavel Matula, Ph.D., Faculty of Informatics, Masaryk University

Distance Measurement in Digital Grids

Abstract: In image analysis, objects of interest are often understood as subsets of regular grids of points (represented, for example, as the foreground of a binary image). This naturally raises the question of how to measure distances between objects in a way that is algorithmically efficient whilst still providing a good approximation of Euclidean distances. In this lecture, we will address precisely this question. In particular, we will present integer approximations of the Euclidean metric in regular digital grids, the computation of a distance map, and demonstrate applications of these approaches.

RNDr. Petr Matula, Ph.D., Faculty of Informatics, Masaryk University

High-content screening from the perspective of image analysis

Abstract: High-content screening (HCS) is a technique based on fully automated optical microscopy used in contemporary biological research and in drug development to identify substances (e.g. chemical compounds, biomolecules, ...) that induce observable (measurable) changes in cells. Fully automated optical microscopes generate such a large volume of data that it is beyond human capacity to examine and evaluate all the captured images. This naturally gives rise to many interesting problems in the field of fully automated image processing (e.g. segmentation, classification, motion tracking, quality control), which need to be addressed. The aim of the lecture is to outline typical HCS challenges from the perspective of image analysis and to present examples of solutions drawn from our own practical experience in collaboration with the German Cancer Research Centre (DKFZ) and the University of Heidelberg.

23 October 2012

Parag Kulkarni, Ph.D., DSc., FI MU

Paradigm of Systemic Machine Learning and Its Role in Knowledge Innovation and Knowledge Discovery

Abstract: The advent of new and powerful tools and technologies for information retrieval has transformed the landscape of information processing. Vast quantities of information in unstructured or semi-structured form, accompanied by verbose comments – and interconnected in profound ways – are now at our disposal. This creates new challenges in the field of knowledge management, machine learning and information processing. Traditional knowledge management and machine learning techniques are no longer relevant and therefore hinder knowledge workers. These techniques typically fail to capture the cadence of organic activity and relevant patterns amongst dependencies, and thus fail to build an overall context and pertinent knowledge. The last decade was all about value innovation, where the strategic significance of activities and initiatives was measured in terms of direct value creation and value delivered. However, value innovation lacks sustainability. Whilst businesses are moving towards knowledge-based foundations with ever-increasing intelligence, there is a growing need for knowledge innovation and systemic machine learning to achieve sustainable value creation. Knowledge innovation is sustainable value creation through a systemic approach and holistic learning. It is based on systemic value creation via systemic learning and the effective use of knowledge. Learning plays a major role in knowledge innovation and, consequently, machine learning can play a key role in knowledge innovation. This talk introduces the innovative concept of Systemic Machine Learning and knowledge innovation, and explores their role, alongside the author’s contributions, in introducing new paradigms in ML and KM and applying them to value creation.

30 October 2012

RNDr. Andriy Stetsko, Faculty of Informatics, Masaryk University

Prof. RNDr. Václav Matyáš, M.Sc., Ph.D., Faculty of Informatics, Masaryk University

Intrusion detection in wireless sensor networks

Abstract: Wireless sensor networks possess a unique combination of characteristics compared to traditional wired and wireless networks. Sensor nodes are highly distributed, often not physically protected and constrained in terms of computational resources (energy, memory, bandwidth and processing power). Due to these constraints, traditional and well-established security mechanisms used in traditional wired and wireless networks cannot be applied (either directly or at all) in wireless sensor networks.

This presentation focuses on a new approach to intrusion detection in wireless sensor networks. We present our framework, which helps to strike a balance between the detection accuracy and resource consumption of an intrusion detection system for a given application scenario. In addition to its conceptual architecture, implementation details and test results, we also cover two topics closely related to our framework, namely the evaluation and simulation of intrusion detection systems for wireless sensor networks.

6 November 2012

Doc. PhDr. Karel Pala, CSc., Faculty of Informatics, Masaryk University

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

New findings in the field of large web corpora

Abstract: In this lecture, we will focus on very large text corpora obtained from the web. The motivation behind the creation of such resources is the need for the largest possible data sets for various types of users and applications. At the level of basic research, the aim is to gain a deeper understanding of the structure of natural language. We will discuss the latest results in this field and mention the techniques developed for building large corpora comprising more than a billion tokens, created recently at the Centre for Natural Language Processing at the Faculty of Informatics, Masaryk University.

Specifically, we will mention the tools that have been developed and are in use, and in particular we will analyse the scalability of the individual technologies employed. We will describe the problems we encounter when processing terabytes of data, as well as their solutions.

13 November 2012

Prof. RNDr. Jiří Zlatuška, CSc., Faculty of Informatics, Masaryk University

Principles of the peer review evaluation procedure for research units in computer science

Abstract: General methodologies for evaluating research and researchers in computer science have distinctive features compared to older and more established scientific disciplines, manifesting themselves in the way new findings are published, whilst taking into account the specific disciplinary characteristics of research in this field, which is not a purely theoretical discipline. This can be demonstrated by comparing the approaches used for evaluating computer science research units abroad. A proposal will be presented evaluation procedure that takes into account the problems and experiences associated with the approaches hitherto used in the Czech Republic, the results of an audit of research and development evaluation in the Czech Republic, and which draws on international practices, including the use of peer review, the principles of which it is desirable to adapt to the domestic context.

20 November 2012

Dr Luděk Smolík, Siegen, Germany

On the diversity of the concept of entropy

Abstract: In 1865, Clausius introduced the concept of entropy into statistical mechanics. During the industrialisation era, this abstract quantity supported thermodynamic calculations for machines converting heat into mechanical work, particularly the steam engine. The Greek word ‘entropy’ means something like a change of form. The definition of entropy allows us to quantitatively monitor the degree of irreversibility of a process. Following the model of statistical mechanics, almost a century later the concept of entropy was pragmatically introduced into the then-new fields of cybernetics and computer science. Today, the concept of entropy has found its way into many areas of our lives. In this lecture, I shall explore entropy using examples of its origins in the 19th century, its development and axiomatic redefinition in the late 20th century, and I shall discuss its appropriate and less appropriate uses.

27 November 2012

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

On the Power of Mathematical Discovery: A Tribute to A.M. Turing on the Centenary of his Birth

Abstract: This year marks the 100th anniversary of the birth of the British mathematician Alan Mathison Turing. Alan Turing’s short and eccentric life had a fundamental influence on the current computer age, the philosophy of mind and the outcome of the Second World War. All this was a consequence of his great mathematical discovery.

In this lecture, we shall review the key milestones of Turing’s life. We will pay particular attention to his mathematical discovery (Turing’s machine), which became the key to the construction of the first computers and initiated the development of new scientific disciplines, such as computability theory, artificial intelligence, computer science and cognitive science. We shall also mention his contribution to the development of numerical mathematics and the theory of morphogenesis. Through his crucial role in cracking the secret code of German submarines, he made a significant contribution to the defeat of Nazism.

4 December 2012

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

Approximate computation in the context of evolutionary design

Abstract: This lecture will introduce evolutionary design as a method for the automatic generation of programmes and computer circuits. It will also explain the concept of approximate computing and the use of evolutionary design for the automatic construction of logic circuits that approximately implement the required function. It will be demonstrated that, if we are able to tolerate inaccuracies in calculations, we can significantly reduce the area and power consumption required to operate the circuit.

11 December 2012

Prof. Ing. Zdeněk Strakoš, DrSc., Faculty of Mathematics and Physics, Charles University, Prague

Challenges in algebraic computations

Abstract: The current state of the art in iterative solvers is the result of tremendous algorithmic developments over the last few decades and of research into how to solve given problems. In this contribution, we focus more on the dual question of why certain approaches do or might not work when solving practical problems. This requires considering the algebraic computations in the context of the mathematical properties of the underlying model and its discretisation, using tools ranging from functional analysis to approximation and matrix theory.

18 December 2012

JUDr. Radim Polčák, Ph.D., Institute of Law and Technology, Faculty of Law, Masaryk University

Cyber security as a legal issue

Abstract: The legal reflection on cyber security is burdened not only by the legislator’s limited awareness of its technical aspects, but also by general confusion regarding fundamental systematic issues. The aim of this paper is to discuss the question of what place specific legislation designed to protect national interests against cyber attacks should occupy within the current legal system, and what role the government cyber security centre should fulfil. Specific issues to be addressed include the constitutional proportionality of the information rights concerned (in comparison with the ill-conceived teleology of the current US legislative initiative), the interpretation of a cyber attack as armed aggression, and the legislative and technical options for implementing protective measures.