Start - Development of Artificial Intelligence methods - Developemnt of machine learning methods

Development of machine learning methods
Cluster analysis
Cluster analysis and data grouping are some of the basic issues of machine learning addressing the problem of unsupervised learning. In this area, our research directions are bi-clustering, i.e. the development of methods combining the problem of data grouping with the feature selection, development of new clustering algorithms, including e.g. linguistic data grouping, enabling linguistic definition of data grouping criteria, and improvement of already existing solutions.
Contact persons (email: firstname.lastname@polsl.pl):
Classification and regression methods
Classification and regression methods are a kind of semi-supervised or supervised methods. In this area, the main challenges we operate-in are explainable machine learning methods such that the decision-making process of the learning systems can be interpreted by humans. This issue also includes feature selection methods and features importance analysis as well as visualization of the extracted knowledge. Another active research area is the automation of machine learning processes (AutoML), including automatic model selection along with the selection of its hyper-parameters, as well as model selection for the ensembles of prediction models. Further work in this subarea focuses on the improvement of existing algorithms and methods such as extending their scalability and increasing prediction accuracy. For example, it results in new architectures of neural renewal. The last issue we are working on are construction of predictive models for complex and non-trivial datasets, including unbalanced data and data streams and multi-labeled data.
Automation of machine learning processes including meta-learning:
- dr hab. inż. Marcin Blachnik, https://orcid.org/0000-0003-3336-4962
- dr inż. Michał Kozielski, https://orcid.org/0000-0003-3573-7638
- dr inż. Daniel Kostrzewa, https://orcid.org/0000-0003-2781-3709
- dr inż. Jakub Nalepa, https://orcid.org/0000-0002-4026-1569
Explainable machine learning methods, and information selection in learning systems
- dr hab. inż. Marek Sikora, prof. PŚ, https://orcid.org/0000-0002-2393-9761
- dr hab. inż. Krzysztof Simiński, prof. PŚ, http://orcid.org/0000-0002-6118-606X
- dr hab. inż. Marcin Blachnik, https://orcid.org/0000-0003-3336-4962
- dr inż. Łukasz Wróbel, https://orcid.org/0000-0003-3573-7638
- dr inż. Adam Gudyś, https://orcid.org/0000-0002-5508-0090
- dr inż. Urszula Stańczyk, https://orcid.org/0000-0002-5071-7187
- dr inż. Sebastian Porębski, https://orcid.org/0000-0001-9926-4265
New architectures of neural networks and new prediction algorithms
- dr hab. inż. Michał Kawulok, prof. PŚ, https://orcid.org/0000-0002-3669-5110
- dr inż. Jakub Nalepa, https://orcid.org/0000-0002-4026-1569
- dr inż. Daniel Kostrzewa, https://orcid.org/0000-0003-2781-3709
- dr hab. inż. Sławomir Golak, prof. PŚ, http://orcid.org/0000-0002-9325-3407
Budowa systemów uczących się dla niestandardowych zestawów danych
- dr inż. Małgorzata Bach, https://orcid.org/0000-0002-6239-7790
- dr inż. Aleksandra Werner, https://orcid.org/0000-0001-6098-0088
- dr inż. Michał Kozielski, https://orcid.org/0000-0003-3573-7638