Date and time:
15 October 2025 at 15:30 pm
Place:
Sano Centre for Computational Medicine
2nd floor, CE building, entrance C5
Czarnowiejska 36, Kraków
Maximum number of participants: 10
Duration: Nh
Jose Sousa1
1Sano Centre for Computational Medicine, Czarnowiejska 36, 30-054 Kraków, Poland
j.sousa@sanoscience.org
Keywords: Explainable AI, Data Abstraction, Knowledge Graphs, Classification, Small Datasets
1. Introduction
This Deep learning has achieved remarkable performance but often requires large datasets, significant computer resources, and lacks transparency, making it hard to trust in sensitive fields such as healthcare and law. The Comprehensive Abstraction and Classification [1–4] Tool for Uncovering Structures (CACTUS) offers a transparent and efficient approach by:
- Supporting small and incomplete datasets,
- Preserving the semantic meaning of categorical variables,
- Building interpretable knowledge graphs for feature interactions,
- Providing feature ranking and community analysis for explainable classification,
- Offering memory-efficient, parallelised analysis.
2. Description of the tutorial
This hands-on session introduces CACTUS for explainable AI and secure analytics. Participants will:
- Learn CACTUS architecture (decision tree, abstraction, correlation modules),
- Prepare datasets and YAML configs for flexible analysis,
- Abstract continuous and categorical features into interpretable forms,
- Generate and interpret knowledge graphs and feature rankings,
- Compare CACTUS with standard ML models on datasets like breast cancer, thyroid, heart disease, and its use on the allergies project.
3. Knowledge and skills to be gained
By the end of this tutorial, participants will be able to:
- Understanding Explainable AI and CACTUS methodology,
- Running CACTUS for binary and multi-class tasks,
- Visualising feature interactions with knowledge graphs,
- Interpreting feature rankings alongside decision trees and correlations,
- Applying best practices for incomplete or small datasets.
Acknowledgements: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 857533 and from the International Research Agendas Programme of the Foundation for Polish Science No MAB PLUS/2019/13. The tutorial was created within the project of the Minister of Science and Higher Education “Support for the activity of Centres of Excellence established in Poland under Horizon 2020” on the basis based on contract number MEiN/2023/DIR/3796. We gratefully acknowledge Poland’s high-performance Infrastructure, PLGrid ACC Cyfronet AGH , for providing computer facilities and support within the computational grant no. PLG/2025/018289.
References:
- Anna Drożdż, Brian Duggan, Mark W. Ruddock, Cherith N. Reid, Mary Jo Kurth, Joanne Watt, Allister Irvine, John Lamont, Peter Fitzgerald, Declan O’Rourke, David Curry, Mark Evans, Ruth Boyd, and Jose Sousa. 2024. Stratifying risk of disease in haematuria patients using machine learning techniques to improve diagnostics. Front. Oncol. 14, (2024), 1401071. https://doi.org/10.3389/fonc.2024.1401071
- Luca Gherardini, Paulina Tworek, Maja Szczypka, Yousef Khan, Marek Mikołajczyk, Roman Lewandowski, and Jose Sousa. 2025. Artificial Intelligence in Medicine, 23rd International Conference, AIME 2025, Pavia, Italy, June 23–26, 2025, Proceedings, Part II. (2025), 171–175. https://doi.org/10.1007/978-3-031-95841-0_32
- Luca Gherardini, Varun Ravi Varma, Karol Capała, Roger Woods, and Jose Sousa. 2024. CACTUS: a Comprehensive Abstraction and Classification Tool for Uncovering Structures. ACM Trans. Intell. Syst. Technol. 15, 3 (2024), 1–23. https://doi.org/10.1145/3649459
- Paulina Tworek, Maja Szczypka, Julia Kahan, Marek Mikołajczyk, Roman Lewandowski, and Jose Sousa. 2025. Artificial Intelligence in Medicine, 23rd International Conference, AIME 2025, Pavia, Italy, June 23–26, 2025, Proceedings, Part I. (2025), 448–456. https://doi.org/10.1007/978-3-031-95838-0_44
