Program 2018/19 – joint Datalys, UXI, NN

Seminars run in three groups separately. In this page we list just joint seminars and other activities for broader audience.

[Datalys 1.31 12:30]
[UXI 2.06 12:45]
[NN 4.08 12:30]

17.4.2018 » IIT.SRC 2019

  • Keynote: Michael Chaudron: The Eyes Are the Windows to the Mind: Implications for Intelligent User Interfaces [presentation] Degree Program PeWe Members

  • ČIEF Matej: Modeling User’s Individual Differences during Reading
  • DRAGÚŇ Dominik: Characterization of Fake News and their Proliferation using Data Analysis
  • DRGOŇA Marek: Aspect Detection for Sentiment Analysis
  • HANÁKOVÁ Martina: Fake News Reading: An Eye-tracking Study
  • LOJA Filip: Web Flashlight
  • MAČINEC Peter: Improving Fake News Detection with Thematically Similar News Groups
  • MAKO Edvin: Interpreting Random Forest Models using Decision Rules
  • PECHER Branislav: Explaining Individual Neural Network Decisions Using Correlated Perturbations
  • BOJKOVSKÝ Michal: Multilingual Hate Speech Detection on Twitter
  • VAŠKO Dominik: Neural Model for Text Generation in Slovak Language
  • ZAŤKO Šimon: Question Popularity Prediction in CQA Systems

Master Degree Program PeWe Members

  • BABINEC Peter: Evaluating the Content Quality of Discussions in MOOC Systems
  • BAKOŠ Peter: Automatic Event Detection in User Interface Video-recordings
  • HUCKO Michal: Identification of User Confusion in a Web Application
  • JANEČEK Jakub: Interpretability of Machine Learning Models Created by Clustering Algorithms
  • KUCHÁR Vladimír: Automatic Text Comprehension Detection
  • MELÚCH Michal: Prediction of Perceived Text Difficulty
  • NAGY Imrich: Processing Sequences of Transactions Using Deep Learning Models
  • PECHER Branislav: Explaining Individual Neural Network Decisions Using Correlated Perturbations
  • REŠUTÍK Lukáš: Detection of Respecting Instruction based on User’s Behavior
  • SAMOTNÝ Ľubomír: Cognitive Style Influence on Individual Web Navigation
  • ŠLESARIKOVÁ Tatiana: Eye Tracking Data Correction Method
  • ŠTEFANCOVÁ Elena: Temporal Recommendations based on Locality-specific Seasonality and Long-term Trends
  • TIBENSKÝ Peter: Context-aware Adaptive Personalized Recommendation: A Meta-hybrid
  • VÍTEK Andrej: Extracting Temporal Eye Gaze Features using Hidden Markov Modelling
  • ZAŤKO Timotej: Cognitive Characteristics and Educational Algorithm Games

Doctoral Degree Program PeWe Members

  • GAŠPAR Peter: Visual Recommendation in Shopping Process
  • HLAVÁČ Patrik: Web-navigation Skill Assessment through Eye-tracking Data
  • JAFARI Pooria: Detecting Extraversion based on Detailed User Activity
  • MOCKO Martin: Utilizing Latent Error Representation for Time Series Prediction
  • PECÁR Samuel: Neural Model Ensemble for Suggestion Mining
  • PIKULIAK Matúš: Combining Multitask and Multilingual Learning with Parameter Sharing
  • RÁC Miroslav: Preference Dynamics and Behavioral Traits in Fashion Domain

TP CUP teams with PeWe Members

  • FABIŠ Michal, RAFČÍKOVÁ Katarína, SITÁROVÁ Daniela, SLANINKA Andrej, VEŠS Maroš, ZAŤKO Andrej, ŽÁK Martin: Webable: Web Browser for Visually Impaired People
  • CÁK Milan, TRAN DUC David, LAM TUAN Dung, BARABÁS Matúš, LIPTÁK Peter, ŽATKOVÁ Veronika, ŽIDULIAK Patrik: MonAnt – Monitoring Online Antisocial Behaviour
  • DOLNÁ Dominika, MALÍK Peter, POHL Tomáš, SCHNEIDER Jozef, ŠČASNÝ Andrej, ŠTEVLÍK Dominik: ImageSearch: Deep Learning based Visual Search for E-commerce
  • GROMA Matej, HORVÁTH Matej, JURKÁČEK Peter, KAMENSKÝ Jozef, KŇAZE Adam, MACKOVÁ Kristína, PEJCHALOVÁ Lenka, SEDLÁŘ Jakub: TrafficWatch: Road Traffic Monitoring Solution for Smart Cities

21.2.2018, 13:00, 2.06 » Habilitation defense and lecture

  • Jakub Šimko: Automatic Assessment of User Performance in Digital Environments

Nowadays, people spend considerable lengths of time working in digital environments of information systems. The primary concern of this habilitation thesis is, how can we automatically measure the quality of human work in these environments? This question is important for multitude of work-like activities. Two specific types of such activities are crowdsourcing processes and user studies. This habilitation thesis examines the existing methods for automatic quality assessment in these domains and summarizes the  contributions of the author in this research field. There are several approach families towards user quality of work assessment. Sometimes, the quality of outputs is directly measured. Another family of approaches infers the quality from user behavior. Other group of approaches is (indirectly) relying on modeling of user skills and abilities. Hence, the approaches examined and presented in this thesis fall into the cross-section of fields of human-computer interaction, user modeling, information systems and machine learning.

13.2.2019, -1.65, 13:45 » PeWe Seminar

  • Ivan Srba, Jakub Šimko: Methodological topics (part 2)
    • How to write good a method/system proposal
    • How to evaluate the contribution of project
    • How to write a good abstract, conclusion and other parts of theses

14.11.2018, -1.58, 12:45 » Seminar

  • Michal Grňo a Viktor Gregor (Pixel Federation): Games, Data and Marketing – on the way to automation

Vo svete mobilných F2P hier je využívanie používateľských dát neoddeliteľnou súčasťou každodenného rozhodovania. Na prednáške sa najprv pozrieme na niekoľko use case-ov data-driven rozhodovania pri nastavovaní stratégie, ale aj pri samotnom vývoji produktov. Prekvapivo najväčším konzumentom dát je marketingové oddelenie. Na čo sú vlastne marketingu dáta, aké dáta zbierame, na čo sa pozeráme, aké KPIs meriame? Ukážeme si, aký prediktívny model v súčasnosti používame na vyhodnocovanie marketingových kampaní, jeho výhody/nevýhody, validácia, hľadanie nových a lepších alternatív.

12.11.2018, -1.65, 16:00 » Data Science Club

  • Dominik Csiba (Innovatrics): Fingerprint recognition: From standard methods to small area matchers
  • Viktor Gregor (Pixel Federation): Faster and better A/B tests with Bayesian inference (video)

5.11.2018, -1.65, 16:00 » Data Science Club

  • Ján Dolinský (Tangent Works): Automatic Model Building for Time-Series with Application in Energy Industry
  • Róbert Magyar (Cellense): Machine Learning in Action – How We Doubled Revenue On A Game With Over A Billion Players

22.10.2018, -1.65, 16:00 » Data Science Club

  • Peter Krátky (Instarea): When relational database is not enough…
  • Ondrej Brichta (Exponea): Events data processing by Spark

8.10.2018, -1.65, 16:00 » Data Science Club

  • Martin Bago (Instarea): How we can quickly find what data we have?
  • Jožo Kováč (Exponea): Let’s investigate the Experience

3.10.2018, -1.58, 12:45 » PeWe Seminar

1.10.2018, -1.65, 16:00 » Data Science Club

  • Dominik Csiba (Innovatrics): From zero to a data science project:

Dominik is a PhD. graduate from the University of Edinburgh, where he focused on the mathematical optimization behind machine learning. Previously he worked at Amazon as a research scientist intern and at Operam as a data scientist, until he settled at Innovatrics on the position of R&D team leader for Bratislava. Additionally, last year he spent some time as a teacher of Calculus at LEAF Academy, pursuing his passion for teaching. Dominik is a very competitive person with a long list of achievements in both math and chess. His current hobby project is the website analyzing public data about Slovak politicians.

From zero to a data science project:
Have you ever wondered how all those data science projects around you looked like at the beginning? What does it take to get tangible results? What technologies do you need to know to even get started? In this talk, we aim to give you some answers to these questions by guiding you through the whole story behind our data science project From initial motivations to our current vision, from a single person to a team of people, from individual goals to general usefulness, we offer you the non-idealized version of our adventure. If you ever considered to start your own data science project and you have a lot of questions about how to do so, we believe this talk might give you the answers you are looking for. We hope to see all of you prospective data scientists there!

  • Renné Donner (contextflow): Julia – a language for fast numerical code as well as general programming

With a background in electrical engineering René has worked for 8 years at the Medical University Vienna as a researcher in computer vision, focussing on anatomical structure localization and content based image retrieval. He is now CTO at contextflow, applying deep learning to large scale medical image data and developing smart tools to aid radiologists in their challenging tasks.

Julia was designed from the beginning for high performance. It is dynamically-typed, feels like a scripting language, and has good support for interactive use. At the same time it compiles to efficient native code for multiple platforms via LLVM.
In this talk we will look at what Julia looks like, what the ecosystem provides and how to start using it in no time.

19.9.2018, -1.58, 12:45 » PeWe Seminar