Ignacy Stepka

PhD Student @ Auton Lab

Machine Learning Department

Carnegie Mellon University

I’m a research assistant at Carnegie Mellon University with Dr. Artur Dubrawski in AutonLab. Right now, I mostly spend time thinking about gradient inversion attacks and privacy issues of federated learning. Before, I spent some time developing adaptive strategies for handling persistent agent failures in decentralized federated learning. I’ve also worked on formal verification of Bayesian Networks and contributed to the DARPA Triage Challenge, developing a Bayesian Network prototype with medical experts for autonomous triage during mass casualty incidents.

In the past, I spent some time as a research assistant at the Machine Learning Laboratory at Poznan University of Technology, collaborating with Prof. Jerzy Stefanowski and Dr. Mateusz Lango on developing robust counterfactual explanations for black-box model explainability.

I was also a part of the R&D team at Poznan Supercomputing and Networking Center (Polish Academy of Sciences), contributing to EU-funded projects involving black-box model analysis and anomaly detection in industrial automotive and HPC environments.

In early 2025 I graduated from Poznan University of Technology with a B.Sc. Eng. in Artificial Intelligence.

news

Jun 06, 2025	Paper on saliency maps and fairness to appear at SIGKDD'2025 (UMC)
Apr 21, 2025	Paper on saliency maps and fairness accepted at XAI-2025
Apr 10, 2025	DetoxAI public release
Nov 17, 2024	Paper accepted at SIGKDD'25 main track
Nov 10, 2024	Three posters at ML in PL in Warsaw, Poland
Sep 13, 2024	Oral presentation at XRAI workshop on ECML PKDD 2024 in Vilnus, Lithuania
May 04, 2024	Poster at GhostDay AMLC in Poznan, Poland
Apr 24, 2023	Poster at 4-th Polish Conference on Artificial Intelligence in Lodz, Poland

selected publications

Conference
Counterfactual Explanations with Probabilistic Guarantees on their Robustness to Model Change

Ignacy Stępka, Mateusz Lango, and Jerzy Stefanowski

In 31st SIGKDD Conference on Knowledge Discovery and Data Mining - Research Track, 2025

Abs DOI arXiv Bib Supp Code Website

Counterfactual explanations (CFEs) guide users on how to adjust inputs to machine learning models to achieve desired outputs. While existing research primarily addresses static scenarios, real-world applications often involve data or model changes, potentially invalidating previously generated CFEs and rendering user-induced input changes ineffective. Current methods addressing this issue often support only specific models or change types, require extensive hyperparameter tuning, or fail to provide probabilistic guarantees on CFE robustness to model changes. This paper proposes a novel approach for generating CFEs that provides probabilistic guarantees for any model and change type, while offering interpretable and easy-to-select hyperparameters. We establish a theoretical framework for probabilistically defining robustness to model change and demonstrate how our BetaRCE method directly stems from it. BetaRCE is a post-hoc method applied alongside a chosen base CFE generation method to enhance the quality of the explanation beyond robustness. It facilitates a transition from the base explanation to a more robust one with user-adjusted probability bounds. Through experimental comparisons with baselines, we show that BetaRCE yields robust, most plausible, and closest to baseline counterfactual explanations.
@inproceedings{stepka2024cfeprob, title = {Counterfactual Explanations with Probabilistic Guarantees on their Robustness to Model Change}, author = {St\k{e}pka, Ignacy and Lango, Mateusz and Stefanowski, Jerzy}, booktitle = {31st SIGKDD Conference on Knowledge Discovery and Data Mining - Research Track}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, year = {2025}, doi = {10.1145/3690624.3709300} }
Conference
DetoxAI: a Python Package for Debiasing Neural Networks

Ignacy Stępka , Lukasz Sztukiewicz, Michał Wiliński, and Jerzy Stefanowski

In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, Porto, Portugal, 2025

Abs Bib Code Website

While machine learning fairness has made significant progress in recent years, most existing solutions focus on tabular data and are poorly suited for vision-based classification tasks, which rely heavily on deep learning. To bridge this gap, we introduce DetoxAI, an open-source Python library for improving fairness in deep learning vision classifiers through post-hoc debiasing. DetoxAI implements state-of-the-art debiasing algorithms, fairness metrics, and visualization tools. It supports debiasing via interventions in internal representations and includes attribution-based visualization tools and quantitative algorithmic fairness metrics to show how bias is mitigated. This paper presents the motivation, design, and use cases of DetoxAI, demonstrating its tangible value to engineers and researchers.
@inproceedings{detoxai2025, title = {DetoxAI: a Python Package for Debiasing Neural Networks}, year = {2025}, author = {St\k{e}pka, Ignacy and Sztukiewicz, Lukasz and Wili\'{n}ski, Micha\l{} and Stefanowski, Jerzy}, address = {Berlin, Heidelberg}, publisher = {Springer-Verlag}, booktitle = {Joint European Conference on Machine Learning and Knowledge Discovery in Databases}, location = {Porto, Portugal}, }