The Foundation for Polish Science announced the results of the first TEAM-NET competition. Co-financing for the total amount of nearly PLN 18 million was received by the consortium CFT, IITiS, UJ.
- Author: Łukasz Pawela | Post date: 19.03.2019
- Author: Zbigniew Puchała | Post date: 11.03.2019 | Last modified: 09.05.2019
Date:27/03/2019 - 13:00
Speaker:Jarosław Duda, Wydział Matematyki i Informatyki, Uniwersytet JagiellońskiWhile machine learning techniques are very powerful, they have some weaknesses, like iterative optimization with many local minimums, large freedom of parameters, lack of their interpretability and accuracy control. From the other side we have classical statistics based on moments not having these issues, but providing only a rough description. I will talk about approach which combines their advantages: with MSE-optimal moment-like coefficients, but designed such that we can directly translate them into probability density. For multivariate case such basis of mixed
- Author: Zbigniew Puchała | Post date: 25.01.2019 | Last modified: 19.02.2019
Date:29/01/2019 - 13:00
Speaker:Ivanna Dronyuk, Lviv Polytechnic National University
A new method for constructing a digital water-marking based on the Ateb-functions is proposed. This new approach belongs to methods of protecting images based on hiding data in the frequency domain. It is a generalization the well-known method using Fourier transform. The proposed method, based on the two dimension Discrete Ateb Transform is resistant to most types of attacks. Our experiments prove its efficiency and stability.
- Author: Jarosław Miszczak | Post date: 21.01.2019 | Last modified: 21.01.2019
Date:31/01/2019 - 13:00
Speaker:Gregorio Ferreira, ArcelorMittal
Machine learning and AI has been a trending topic for many years already. However, it is well known the challenges companies are facing to deploy robust solutions taking advantage of this kind of algorithms. The gap between R&D and IT-organisation is still latent, and until both of them don't get to a common point in which the cost and complexity of deploying AI-based solutions are well understood, the possibilities to extract all the value from data are diminished.
- Author: Jarosław Miszczak | Post date: 11.01.2019 | Last modified: 29.01.2019
Date:30/01/2019 - 12:00
Speaker:Małgorzata Domino, SGGW w Warszawie, Centrum Badań Biomedycznych
We call the bioelectric signal an electrical signal that arises in the tissues of living organisms. In particular, the signal generated in the complex functional system of the smooth muscle is interesting and described using EMG. The electromyogram reflects the bioelectrical activity of the smooth muscle of the reproductive system in the form of bioelectric activity recording preceding each systolic activity. At the same time, not all EMG activity triggers systolic activity.
- Author: Piotr Gawron | Post date: 17.07.2018
IITiS PAN employee Mateusz Ostaszewski received a scholarship in the Etiuda 6 competition of the National Science Center for the project entitled "The use of machine learning in quantum information". The project will be implemented in cooperation with the Department of Computer Science, University College London.
- Author: Piotr Gawron | Post date: 04.07.2018
Our employee and student of the Faculty of Applied Mathematics at the Silesian University of Technology, Bartosz Grabowski, received the Diamond Grant for the project entitled "Application of knowledge transfer for convolutional neural networks to improve the classification of hyperspectral images".
- Author: Łukasz Pawela | Post date: 14.06.2018 | Last modified: 18.06.2018
Date:27/06/2018 - 12:15
Speaker:Hanna Wojewódka, Uniwersytet Śląski
The central limit theorem (CLT) and the law of the iterated logarithm (LIL) are, along the strong law of large numbers (SLLN), the most common limit theorems. Some well-known results concerning limit theorems, obtained mainly due to the martingale method, are gathered in . Although the asymptotic behaviour of stationary and ergodic Markov chains is already well investigated, limit theorems for a wider class of Markov processes are still the subject of research. Together with D. Czapla and K.