Machine Learning Group

The Machine Learning Group specializes in the design and selection of machine learning algorithms and models for application projects, primarily those related to computer vision and time series. The team also conducts basic research in the field of machine learning methods for hyperspectral imaging, and recently has focused primarily on deep learning—specifically, issues related to “dead neurons,” continuous learning, and reinforcement learning. Originating from the Multimedia Systems Group, the team members have extensive experience in data analysis and mining (including hyperspectral images, biomedical data, 3D images, and signal data), processing methods (including statistical classifiers and deep learning architectures), as well as the technical and organizational aspects of carrying out research and implementation projects (including the process of preparing and supporting the implementation of machine learning systems for specific problems).

Current Projects:

1. WaterPrime (2021–), No. POIR.01.01.01-00-1414/20, a project aimed at developing and implementing a smart data analytics platform to detect leaks and monitor the condition of water supply networks. The project is in the pilot launch phase (3 water utilities, 20+ water supply zones, approx. 10,000 monitored devices), and even at this stage, it has already contributed to the detection of a number of failures and a significant reduction in water losses (more information at https://www.iitis.pl/pl/project/ekosystem-intelligence-augmentation-dla-analityk%C3%B3w-sieci-dystrybucji-wody, https://aiut.com/en/waterprime-artificial-intelligence-to-help-cities-detect-water-leaks/, https://doi.org/10.2166/ws.2023.118)
2. Supporting the diagnosis of selected diseases using biomarkers, a project carried out in collaboration with the Department of Psychiatry at the Medical University of Silesia in Tarnowskie Góry. The goal of the project is to conduct pilot studies of a parametric method for assessing the severity of symptoms of, among others, schizophrenia and bipolar disorder using HRV signals and accelerometers, in order to support medical diagnosis (more information at https://doi.org/10.1101/2023.08.04.23293640, https://zenodo.org/records/8171266)

Key topics addressed in the past:

1. Active Shape Network—a series of projects focused on the use of probabilistic graphical models for processing hyperspectral imagery (more information at https://doi.org/10.1016/j.forsciint.2021.110701, https://doi.org/10.3390/rs12162653, https://doi.org/10.1016/j.culher.2018.01.003, https://doi.org/10.1016/j.isprsjprs.2016.08.011)
2. A series of implementation projects related to the design of machine learning components for diagnosing problems in liquid fuel distribution systems (more information at https://www.iitis.pl/pl/project/system-gromadzenia-i-analizy-danych-o-charakterze-strumieniowym-dedykowanego-dla-sieci, https://www.iitis.pl/pl/project/badanie-i-rozw-j-wdro-e-demonstracyjnych-inteligentnego-systemu-zarz-dzania-stanami-paliw-i, https://www.iitis.pl/en/project/opracowanie-i-budowa-modeli-oraz-metod-implementacji-inteligentnych-system%C3%B3w-monitorowania)
3. A series of projects related to image processing, including industrial inspection and support for UAV operators. In the latter case, the project titled “Stabilization and Tracking Module Developed for the FlyEye Unmanned Aerial Vehicle (UAV) System Manufactured by Flytronic sp. z o.o.” aimed to develop algorithms and implement image processing components for stabilization and tracking tasks specified by the UAV operator. A demonstration of the FlyEye system received an honorable mention from the Minister of Internal Affairs at the 2010 International Defense Industry Exhibition in Kielce.

(Complete list of publications and projects - https://www.iitis.pl/person/pglomb, https://www.iitis.pl/pl/research-group/zesp%C3%B3%C5%82-uczenia-maszynowego)

Machine-translated
Przetłumaczone maszynowo