Research
Science behind ValViser
Behind ValViser is research from the Idiap Research Institute’s Neuro-symbolic AI Lab on turning complex operational data into decisions: predictive models of how machines are used and perform, AI systems that reason with expert knowledge, and human–AI interaction for trustworthy decision support. ValViser brings that work from the lab into industry as an Idiap spin-off. Use the graph to explore the papers and themes — or open Google Scholar for the full trail.
January 2026
Encoding expert rules with LLMs
Franciszek Górski, Oskar Wysocki, Marco Valentino, André Freitas, Andrzej Czyżewski
IEEE Access
Read paperOctober 2025
Multi-agent evidence you can audit
Oskar Wysocki, Magdalena Wysocka, M. Jacobo, Harriet Unsworth, André Freitas
arXiv preprint
Read paperOctober 2025
How people perceive AI-assisted systems
A. A. Andhale, Sammie Mak, M. Lee, Oskar Wysocki, S. Singh, Harriet Unsworth, Donna M. Graham, et al.
Annals of Oncology (ESMO Congress 2025)
Read paperSeptember 2025
Making AI outputs make sense to operators
Oskar Wysocki, Sammie Mak, Hannah Frost, Donna M. Graham, Dónal Landers, Tariq Aslam
International Journal of Medical Informatics
Read paperJuly 2025
Digital twins for urban fleets
Oskar Wysocki, Michał Kuziemski, André Freitas, Michał Wasilczuk, Jacek Czyżewicz
IEEE Access
Read paperApril 2025
Stress-testing reasoning in LLMs
Magdalena Wysocka, Danilo S. Carvalho, Oskar Wysocki, Marco Valentino, André Freitas
NAACL 2025
Read paperOctober 2024
Keeping AI answers grounded in facts
Magdalena Wysocka, Oskar Wysocki, Maxime Delmas, Victor Mutel, André Freitas
Journal of Biomedical Informatics
Read paperAugust 2024
Synthesizing knowledge into decisions
Oskar Wysocki, Magdalena Wysocka, Danilo Carvalho, Alex Bogatu, D. M. Gusicuma, Maxime Delmas, et al.
ACL 2024 (System Demonstrations)
Read paperMay 2023
ML guided by structured evidence
Alex Bogatu, Magdalena Wysocka, Oskar Wysocki, H. Butterworth, M. Pillai, et al.
Journal of Biomedical Informatics
Read paperMay 2023
Models shaped by domain structure
Magdalena Wysocka, Oskar Wysocki, M. Zufferey, Dónal Landers, André Freitas
BMC Bioinformatics
Read paperMarch 2023
Seeing the argument behind a decision
H. Mardah, Oskar Wysocki, Markel Vigo, André Freitas
arXiv preprint
Read paperMarch 2023
When experts trust AI decisions
Oskar Wysocki, Jack K. Davies, Markel Vigo, Anne C. Armstrong, Dónal Landers, Rebecca Lee, André Freitas
Artificial Intelligence
Read paperMarch 2023
What models need to know about a domain
Oskar Wysocki, Zili Zhou, Paul O’Regan, Daniel Ferreira, Magdalena Wysocka, Dónal Landers, et al.
Computational Linguistics
Read paperAugust 2022
Scoring risk when every signal matters
Oskar Wysocki, Cong Zhou, Jacobo Rogado, Prerana Huddar, Rohan Shotton, et al.
Cancers
Read paperMay 2022
From messy data to a risk tool
Rebecca J. Lee, Oskar Wysocki, Cong Zhou, Rohan Shotton, et al.
JCO Clinical Cancer Informatics
Read paperSeptember 2019
Modeling heavy-duty fuel use from real data
Oskar Wysocki, Lipika Deka, David Elizondo, Jacek Kropiwnicki, Jacek Czyżewicz
ICONAC 2019
Read paperOctober 2018
Measuring real-world duty-cycle efficiency
Oskar Wysocki, Tomasz Zajdziński, Jacek Czyżewicz, Jacek Kropiwnicki
IOP Conference Series: Materials Science and Engineering
Read paperOctober 2018
Finding efficient engine operating regions
Oskar Wysocki, Jacek Kropiwnicki, Tomasz Zajdziński, Jacek Czyżewicz
IOP Conference Series: Materials Science and Engineering
Read paperOctober 2018
Energetics of machine hydraulic systems
Tomasz Zajdziński, Oskar Wysocki, Jacek Czyżewicz, Jacek Kropiwnicki
IOP Conference Series: Materials Science and Engineering
Read paperSeptember 2018
Working-cycle efficiency from exploitation data
Oskar Wysocki, Tomasz Zajdziński, Jacek Czyżewicz
AUTOBUSY – Technika, Eksploatacja, Systemy Transportowe
Read paperDecember 2017
Engine maps from operational telematics
Oskar Wysocki, Jacek Kropiwnicki, Jacek Czyżewicz
Combustion Engines
Read paper