Spring 2020 Program

This is the programme of the Sitola seminar in spring 2020. Presentations for the current semester are available here.

• 19.2.2020
  Spring holidays (no presentation)
• 26.2.2020
  RNDr. Martin Ukrop
  Let’s question asking questions
  Do you ask questions in the class? Do students answer them? What can you do if the question is followed by silence? Do you learn something from the answers you get? Can questions help the student without giving out any information? Can questions be useful even if you don’t understand the answers? That’s a lot of questions! In this interactive workshop, we’ll be looking at questions, their variants and intentions behind them (and answer some of those posed in this abstract), You’ll be asked questions and you’ll be asking questions, thinking about them and analyzing the effects they may have.
  This session may give you a glimpse of the content discussed at the Teaching Lab, a local course/community aimed to improve teaching at the faculty.
• 4.3.2020
  Mgr. Marian Pavelka, Ph.D.
  CzechGlobe – Change is a challenge
  Abstract: CzechGlobe – Global Change Research Institute of the Czech Academy of Sciences (GCRI) is a public research institution, the European center of excellence investigating the ongoing global change and its impact on the atmosphere, biosphere, and human society through the use of the latest techniques and instrumentation. The research focuses primarily on the development of climate and its future scenarios on the carbon cycle and the effects of changing conditions on the production and biodiversity of ecosystems and on the impacts on the future development and behavior of our society. The integral part of the CzechGlobe are activities aiming at the development of technological innovation processes, proposals of measures for adaptation, and educational activities.
• 11.3.2020 Canceled due to due to coronavirus
  RNDr. Jiří Filipovič, Ph.D.
  Kernel Tuning Toolkit: A framework for dynamic autotuning of highly-efficient CUDA and OpenCL kernels
• 18.3.2020 Canceled due to due to coronavirus
  ing. Michal Bidlo, Ph.D.
  Advances in evolutionary design of cellular automata
• 25.3.2020
  Mgr. Aleš Křenek, Ph.D.
  Containers, scientists, and computing centres – good frenemies
  Abstract: Containers, a lightweight virtualization instrument, were invented (mostly) to isolate pieces of software that may interfere, and their usage is becoming an industry standard.
  Scientists tend to use many pieces of software together in a messy way, bringing eligible use cases to be containerized. But they rarely possess time and skills to deliver such solutions. Some of them try, but they may end up creating all-in-one many-gigabyte containers, contradicting the original container design idea and losing many of its advantages.
  And many old-school system administrators of computing centers hate containers rather openly, being paranoidly scared of potential security risks, and also jealous of users maintaining their software on their own sometimes.
  In this talk, I will follow a real-world scientific use case. I will outline the complexity of its software dependencies and will present its implementation in a Jupyter notebook, containerized itself, and using side containers with large software packages as a reasonable trade-off design. The approach can be easily used on the user’s laptop or moderately powerful desktop machine for flexible development with Docker. Options to transfer the resulting application to a computing center for large scale computation, replacing Docker with less aggressive alternatives (Singularity and Podman) will be discussed finally.
  Slides and examples from the presentation
• 1.4.2020
  Amin Nezarat, Ph.D.
  Big Data and HPC convergence era and our interests
  Abstract: After 50+ years of efforts in HPC, the scientists who are active in this field، introducing the new concepts of the ExaScale era. In this era, we are facing big issues with huge amounts of data. The only way that we can rely on it for solving such problems is Big Data methods and technologies. When Doug Cutting inspired by Google’s paper and MPI project for Hadoop developing, at the same time, computational scientists were developing new technology such as HDF5, Ceph. And now, new mega projects such as ECP or one part of H2020 are proposing the new concepts in this era with convergence ideas (Kokos, PGAS, DAAL…). As a data scientist, I have focused on this field and am trying to converge these concepts for solving new appeared problems in other disciplines. I try to Use HPC + Big data concepts for proposing new methods in my research. In this presentation, I will do my best to familiarize you with my previous research in Bioinformatics, Machine Learning, and data warehouse.
• 8.4.2020
  RNDr. Vít Rusňák, Ph.D.
  Visualizations in the R&D Projects of CSIRT-MU
  Abstract: The Cybersecurity Team of Masaryk University (CSIRT-MU) not only protects and maintains a safe cyber-environment but is also very active in various research and development projects and collaborates with many institutions and companies from public and private sectors. I am a member of CSIRT-MU since 2018. In my work, I focus on research and development of user interfaces and interactive visualizations in the cybersecurity context. In the past two years, we have completely re-designed the user interface of KYPO Cyber Range – a platform for cybersecurity education, created Security Dashboard – a tool for sharing various cybersecurity-related information with the IT community at MU and prototyped new visualizations for Axenta and Flowmon Networks. In the talk, I will overview some of these results and outline our future work. I will also share my views on the state of cybersecurity tools, the importance of domain knowledge, the value of the user-centered approach for the design of new tools, and the need for evaluation methods from human-computer interaction when doing – not only cybersecurity – visualizations research.
• 15.4.2020
  Mgr. Ing. Jiří Marek
  Open Science – Current status and emerging trends
  Open access to scientific information (also known as Open Science) as a new paradigm for contemporary scientific practice is growing rapidly. Higher availability of scientific information improves public services and enables innovation to be introduced more quickly and more efficiently. The lecture will present definitions, trends, and current status of development of Open Access in the local, but also in a broader national and international context. The second part will be focused on the topic of opening research data as an emerging concept of Open Science with Data Management Plans and FAIR principles (Findable-Accessible-Interoperable-Reusable) as key enablers of this transition.
  Slides in PDF
• 22.4.2020
  RNDr. Jiří Filipovič, Ph.D.
  Kernel Tuning Toolkit: A framework for dynamic autotuning of highly-efficient CUDA and OpenCL kernels
  Abstract: In recent years, the heterogeneity of both commodity and supercomputer hardware has increased sharply. Accelerators, such as GPUs or Intel Xeon Phi co-processors, are often essential to improving speed and energy efficiency of highly-parallel codes. However, due to the complexity of heterogeneous architectures, optimization of codes for a certain type of architecture as well as porting codes across different architectures, while maintaining a comparable level of performance, can be extremely challenging. Addressing the challenges associated with performance optimization and performance portability, autotuning has gained much interest. Autotuning of performance-relevant source-code parameters allows us to automatically tune applications without hard coding optimizations and thus helps with keeping the performance portable.
  In this talk, we introduce our Kernel Tuning Toolkit framework and show using a benchmark set that autotuning allows us to reach near-peak performance on various GPUs and outperform baseline implementations on CPUs and Xeon Phis. In addition to offline tuning, we also introduce dynamic autotuning of code optimization parameters during application runtime. With dynamic tuning, the Kernel Tuning Toolkit enables applications to re-tune performance-critical kernels at runtime whenever needed, for example, when input data changes. Although it is generally believed that autotuning spaces tend to be too large to be searched during application runtime, we show that it is not necessarily the case when tuning spaces are designed rationally. Finally, we demonstrate how dynamic performance tuning can be integrated into a real-world application from cryo-electron microscopy domain.
• 29.4.2020
  RNDr. Pavel Troubil, Ph.D.
  Unexpected events on the railway: fast replanning of fleet
  On the railway, many unexpected events happen daily: trains are delayed, tracks become blocked by a tree or a breakdown, and unfortunately, also accidents occur. Railway operators, as well as transportation providers, need to respond by swift replanning of their operations. Train services need to be rescheduled in time and on tracks; fleet needs to cover demands the updated services, potentially replacing unavailable train sets; and finally, the crew needs to be reassigned to operate the vehicles.
  The talk will focus on the replanning of the fleet upon a disturbance. Our approach needs just a couple minutes to provide a new plan while optimizing the number of train sets used and operational costs. Our algorithm minimizes changes to the regular plan and adheres to scheduled maintenance of rolling stock.
• 6.5.2020
  RNDr. Jan Sedmidubsky, Ph.D.
  Intelligent management of human motion data
  Abstract: Human motion can be described by a sequence of skeleton poses, where each pose keeps 2D or 3D coordinates of important body joints in a specific time moment. Such spatio-temporal data can be acquired using dedicated hardware technologies or recent pose-estimation methods capable of determining 2D or even 3D joint positions from ordinary videos. The acquired data have enormous application potential in many fields, ranging from computer animation, security, and sports, to medicine. To make the recorded data more valuable and reusable, intelligent management techniques are required. The objective of this presentation is to introduce basic techniques for classification, annotation, and searching in human motion data. The presented techniques are usually based on a combination of deep-learning and similarity-search principles. Selected techniques will also be demonstrated by interactive web applications, developed by the members of the DISA laboratory.
• 13.5.2020
  Discussion about future Sitola
• 20.5.2020
  ing. Michal Bidlo, Ph.D.
  Advances in evolutionary design of cellular automata
  Abstract: The talk will be devoted to the latest research results regarding an evolutionary design of complex multi-state cellular automata (CA). In particular, two-dimensional cellular automata will be considered in combination with a pattern development problem as a case study. An advanced representation technique for designing transitions functions for CA, called Conditionally Matching Rules, will be briefly introduced and its abilities compared to the conventional table-based representation. It will be demonstrated that the representation significantly influences both the quality of results and the way of their functioning. Interactive visualizations of selected results will also be included.
• 27.5.2020
  Discussion about future Sitola (continuation)
• Examination period 25.5. – 30.9.2020
  Meetings devoted to state exam rehearsals below are scheduled mostly. All state exam rehearsals are in the Czech language. More days can be added if needed by students.
  
  
• 17.6.2020
  Rehearsal of state exam defense (bachelor)
  Vladimír Višňovský: Workflow výpočtu korekce silového pole malých molekul
  Supervisor: Aleš Křenek
  Reader: Miroslav Ruda

  Ivana Krumlová: Nástroj pro sběr a analýzu dat z autoritativních DNS serverů
  Supervisor: Miloš Liška
  Reader: Tomáš Rebok

• 2.9.2020, videoconference (non-standard URL distributed by mail)
  Rehearsal of state exam defense
  Ján Štefkovič : Inovativní přístup k realizaci výpočtů na superpočítačových infrastrukturách
  Supervisor: Tomáš Rebok
  Reader: Lukáš Hejtmánek

• 9.9.2020 2pm (online, one hour earlier!)
  Rehearsal of state exam defense
  Alexandra Nyitraiová: Resource scheduling in OpenStack
  Supervisor: Hana Rudová
  Reader: Dalibor Klusáček

  Petr Valenta: Dynamic vehicle routing problems
  Supervisor: Hana Rudová
  Reader: Pavel Troubil
  

• 16.9.2020, online
  Rehearsal of state exam defense
  Erik Moravec: Kalkulačka možných částek poplatku za prodlouženou dobu studia
  Supervisor: Lukáš Ručka
  Reader: Vladimír Štill

  Filip Müller: Evideční systém vyšetření magnetické rezonance
  Supervisor: Aleš Křenek
  Reader: Jan Fousek
  

Contact: Hana Rudová