Jerzy Stefanowski

Riccardo Albertoni, Sara Colantonio, Piotr Skrzypczyński et al.

Reproducibility is one of the core dimensions that concur to deliver Trustworthy Artificial Intelligence. Broadly speaking, reproducibility can be defined as the possibility to reproduce the same or a similar experiment or method, thereby obtaining the same or similar results as the original scientists. It is an essential ingredient of the scientific method and crucial for gaining trust in relevant claims. A reproducibility crisis has been recently acknowledged by scientists and this seems to affect even more Artificial Intelligence and Machine Learning, due to the complexity of the models at the core of their recent successes. Notwithstanding the recent debate on Artificial Intelligence reproducibility, its practical implementation is still insufficient, also because many technical issues are overlooked. In this survey, we critically review the current literature on the topic and highlight the open issues. Our contribution is three-fold. We propose a concise terminological review of the terms coming into play. We collect and systematize existing recommendations for achieving reproducibility, putting forth the means to comply with them. We identify key elements often overlooked in modern Machine Learning and provide novel recommendations for them. We further specialize these for two critical application domains, namely the biomedical and physical artificial intelligence fields.

Agnieszka Lipska, Jerzy Stefanowski

This work is aimed at the experimental studying the influence of local data characteristics and drifts on the difficulties of learning various online classifiers from multi-class imbalanced data streams. Firstly we present a categorization of these data factors and drifts in the context of imbalanced streams, then we introduce the generators of synthetic streams that model these factors and drifts. The results of many experiments with synthetically generated data streams have shown a much greater role of the overlapping between many minority classes (the type of borderline examples) than for streams with one minority class. The presence of rare examples in the stream is the most difficult single factor. The local drift of splitting minority classes is the third influential factor. Unlike binary streams, the specialized UOB and OOB classifiers perform well enough for even high imbalance ratios. The most challenging for all classifiers are complex scenarios integrating the drifts of the identified factors simultaneously, which worsen the evaluation measures in the case of a several minority classes stronger than for binary ones. This is an extended version of the short paper presented at LIDTA'2022 workshop at ECMLPKDD2022.

Oleksii Furman, Patryk Wielopolski, Łukasz Lenkiewicz et al.

The growing complexity of AI systems has intensified the need for transparency through Explainable AI (XAI). Counterfactual explanations (CFs) offer actionable "what-if" scenarios on three levels: Local CFs providing instance-specific insights, Global CFs addressing broader trends, and Group-wise CFs (GWCFs) striking a balance and revealing patterns within cohesive groups. Despite the availability of methods for each granularity level, the field lacks a unified method that integrates these complementary approaches. We address this limitation by proposing a gradient-based optimization method for differentiable models that generates Local, Global, and Group-wise Counterfactual Explanations in a unified manner. We especially enhance GWCF generation by combining instance grouping and counterfactual generation into a single efficient process, replacing traditional two-step methods. Moreover, to ensure trustworthiness, we innovatively introduce the integration of plausibility criteria into the GWCF domain, making explanations both valid and realistic. Our results demonstrate the method's effectiveness in balancing validity, proximity, and plausibility while optimizing group granularity, with practical utility validated through practical use cases.

Witold Andrzejewski, Jedrzej Potoniec, Maciej Drozdowski et al.

In this paper, we consider energy demand prediction in district heating systems. Effective energy demand prediction is essential in combined heat power systems when offering electrical energy in competitive electricity markets. To address this problem, we propose two sets of algorithms: (1) a novel extension to the algorithm proposed by E. Dotzauer and (2) an autoregressive predictor based on hour-of-week adjusted linear regression on moving averages of energy consumption. These two methods are compared against state-of-the-art artificial neural networks. Energy demand predictor algorithms have various computational costs and prediction quality. While prediction quality is a widely used measure of predictor superiority, computational costs are less frequently analyzed and their impact is not so extensively studied. When predictor algorithms are constantly updated using new data, some computationally expensive forecasting methods may become inapplicable. The computational costs can be split into training and execution parts. The execution part is the cost paid when the already trained algorithm is applied to predict something. In this paper, we evaluate the above methods with respect to the quality and computational costs, both in the training and in the execution. The comparison is conducted on a real-world dataset from a district heating system in the northwest part of Poland.

Ignacy Stępka, Mateusz Lango, Jerzy Stefanowski

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.

Jacek Karolczak, Jerzy Stefanowski

Prototype-based explanations offer an intuitive, example-based approach to support the interpretability of machine learning black box classifiers but often lack feature-level granularity. We introduce a framework that integrates feature importance at two levels to address this gap. First, for local explanations, we propose \textit{alike parts}: a method that uses feature importance scores to highlight the most relevant, shared feature subsets between a classified instance and its nearest prototype, guiding user attention. Second, we augment the global prototype selection objective function with a feature importance term to actively promote diversity in the feature attributions of the selected prototypes. Experiments on six benchmark datasets show that this augmented selection process maintains or, in some cases, increases the prediction fidelity of the surrogate model, suggesting that feature diversity does not compromise model fidelity.

Natalia Wojak-Strzelecka, Szymon Bobek, Grzegorz J. Nalepa et al.

Anomaly and failure detection methods are crucial in identifying deviations from normal system operational conditions, which allows for actions to be taken in advance, usually preventing more serious damages. Long-lasting deviations indicate failures, while sudden, isolated changes in the data indicate anomalies. However, in many practical applications, changes in the data do not always represent abnormal system states. Such changes may be recognized incorrectly as failures, while being a normal evolution of the system, e.g. referring to characteristics of starting the processing of a new product, i.e. realizing a domain shift. Therefore, distinguishing between failures and such ''healthy'' changes in data distribution is critical to ensure the practical robustness of the system. In this paper, we propose a method that not only detects changes in the data distribution and anomalies but also allows us to distinguish between failures and normal domain shifts inherent to a given process. The proposed method consists of a modified Page-Hinkley changepoint detector for identification of the domain shift and possible failures and supervised domain-adaptation-based algorithms for fast, online anomaly detection. These two are coupled with an explainable artificial intelligence (XAI) component that aims at helping the human operator to finally differentiate between domain shifts and failures. The method is illustrated by an experiment on a data stream from the steel factory.

Marcin Kostrzewa, Jerzy Stefanowski, Maciej Zięba

Counterfactual explanations (CFEs) provide actionable recourse, but most methods assume a static framework with fixed data and a trained classifier. This assumption breaks in evolving data environments, such as data streams, where online models are repeatedly updated under concept drift. We identify CFE maintenance in this setting as a previously overlooked problem: explanations that are valid when generated may silently become invalid as the model evolves, including robust CFEs, which are not designed for continuous drift. We propose a lightweight, model-agnostic update scheme that repairs existing CFEs using local sampling to estimate validity and plausibility directions while preserving proximity to the original instance. Experiments on synthetic drifting streams show that initially created CFEs rapidly lose validity, whereas maintained CFEs preserve validity and local plausibility at a lower cost than repeated regeneration.

Ignacy Stępka, Lukasz Sztukiewicz, Michał Wiliński et al.

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.

Marcin Kostrzewa, Maciej Zięba, Jerzy Stefanowski

Counterfactual explanations (CFEs) are essential for interpreting black-box models, yet they often become invalid when models are slightly changed. Existing methods for generating robust CFEs are often limited to specific types of models, require costly tuning, or inflexible robustness controls. We propose a novel approach that jointly models the data distribution and the space of plausible model decisions to ensure robustness to model changes. Using a probabilistic consensus over a model ensemble, we train a conditional normalizing flow that captures the data density under varying levels of classifier agreement. At inference time, a single interpretable parameter controls the robustness level; it specifies the minimum fraction of models that should agree on the target class without retraining the generative model. Our method effectively pushes CFEs toward regions that are both plausible and stable across model changes. Experimental results demonstrate that our approach achieves superior empirical robustness while also maintaining good performance across other evaluation measures.

Jakub Raczyński, Mateusz Lango, Jerzy Stefanowski

Providing natural language explanations for recommendations is particularly useful from the perspective of a non-expert user. Although several methods for providing such explanations have recently been proposed, we argue that an important aspect of explanation quality has been overlooked in their experimental evaluation. Specifically, the coherence between generated text and predicted rating, which is a necessary condition for an explanation to be useful, is not properly captured by currently used evaluation measures. In this paper, we highlight the issue of explanation and prediction coherence by 1) presenting results from a manual verification of explanations generated by one of the state-of-the-art approaches 2) proposing a method of automatic coherence evaluation 3) introducing a new transformer-based method that aims to produce more coherent explanations than the state-of-the-art approaches 4) performing an experimental evaluation which demonstrates that this method significantly improves the explanation coherence without affecting the other aspects of recommendation performance.

Dariusz Brzezinski, Julia Stachowiak, Jerzy Stefanowski et al.

Society is increasingly relying on predictive models in fields like criminal justice, credit risk management, or hiring. To prevent such automated systems from discriminating against people belonging to certain groups, fairness measures have become a crucial component in socially relevant applications of machine learning. However, existing fairness measures have been designed to assess the bias between predictions for protected groups without considering the imbalance in the classes of the target variable. Current research on the potential effect of class imbalance on fairness focuses on practical applications rather than dataset-independent measure properties. In this paper, we study the general properties of fairness measures for changing class and protected group proportions. For this purpose, we analyze the probability mass functions of six of the most popular group fairness measures. We also measure how the probability of achieving perfect fairness changes for varying class imbalance ratios. Moreover, we relate the dataset-independent properties of fairness measures described in this paper to classifier fairness in real-life tasks. Our results show that measures such as Equal Opportunity and Positive Predictive Parity are more sensitive to changes in class imbalance than Accuracy Equality. These findings can help guide researchers and practitioners in choosing the most appropriate fairness measures for their classification problems.

Salvatore Greco, Jacek Karolczak, Roman Słowiński et al.

Explainable artificial intelligence (XAI) has predominantly focused on generating model-centric explanations that approximate the behavior of black-box models. However, such explanations often overlook a fundamental aspect of interpretability: different users require different explanations depending on their goals, preferences, and cognitive constraints. Although recent work has explored user-centric and personalized explanations, most existing approaches rely on heuristic adaptations or implicit user modeling, lacking a principled framework for representing and learning individual preferences. In this paper, we consider Preference-Based Explainable Artificial Intelligence (PREF-XAI), a novel perspective that reframes explanation as a preference-driven decision problem. Within PREF-XAI, explanations are not treated as fixed outputs, but as alternatives to be evaluated and selected according to user-specific criteria. In the PREF-XAI perspective, here we propose a methodology that combines rule-based explanations with formal preference learning. User preferences are elicited through a ranking of a small set of candidate explanations and modeled via an additive utility function inferred using robust ordinal regression. Experimental results on real-world datasets show that PREF-XAI can accurately reconstruct user preferences from limited feedback, identify highly relevant explanations, and discover novel explanatory rules not initially considered by the user. Beyond the proposed methodology, this work establishes a connection between XAI and preference learning, opening new directions for interactive and adaptive explanation systems.

Ignacy Stępka, Mateusz Lango, Jerzy Stefanowski

Counterfactuals are widely used to explain ML model predictions by providing alternative scenarios for obtaining the more desired predictions. They can be generated by a variety of methods that optimize different, sometimes conflicting, quality measures and produce quite different solutions. However, choosing the most appropriate explanation method and one of the generated counterfactuals is not an easy task. Instead of forcing the user to test many different explanation methods and analysing conflicting solutions, in this paper, we propose to use a multi-stage ensemble approach that will select single counterfactual based on the multiple-criteria analysis. It offers a compromise solution that scores well on several popular quality measures. This approach exploits the dominance relation and the ideal point decision aid method, which selects one counterfactual from the Pareto front. The conducted experiments demonstrated that the proposed approach generates fully actionable counterfactuals with attractive compromise values of the considered quality measures.

Damian Horna, Lango Mateusz, Jerzy Stefanowski

Improving the classification of multi-class imbalanced data is more difficult than its two-class counterpart. In this paper, we use deep neural networks to train new representations of tabular multi-class data. Unlike the typically developed re-sampling pre-processing methods, our proposal modifies the distribution of features, i.e. the positions of examples in the learned embedded representation, and it does not modify the class sizes. To learn such embedded representations we introduced various definitions of triplet loss functions: the simplest one uses weights related to the degree of class imbalance, while the next proposals are intended for more complex distributions of examples and aim to generate a safe neighborhood of minority examples. Similarly to the resampling approaches, after applying such preprocessing, different classifiers can be trained on new representations. Experiments with popular multi-class imbalanced benchmark data sets and three classifiers showed the advantage of the proposed approach over popular pre-processing methods as well as basic versions of neural networks with classical loss function formulations.

Jacek Karolczak, Jerzy Stefanowski

Although prototype-based explanations provide a human-understandable way of representing model predictions they often fail to direct user attention to the most relevant features. We propose a novel approach to identify the most informative features within prototypes, termed alike parts. Using feature importance scores derived from an agnostic explanation method, it emphasizes the most relevant overlapping features between an instance and its nearest prototype. Furthermore, the feature importance score is incorporated into the objective function of the prototype selection algorithms to promote global prototypes diversity. Through experiments on six benchmark datasets, we demonstrate that the proposed approach improves user comprehension while maintaining or even increasing predictive accuracy.

Jacek Karolczak, Jerzy Stefanowski

The ability to interpret machine learning model decisions is critical in such domains as healthcare, where trust in model predictions is as important as their accuracy. Inspired by the development of prototype parts-based deep neural networks in computer vision, we propose a new model for tabular data, specifically tailored to medical records, that requires discretization of diagnostic result norms. Unlike the original vision models that rely on the spatial structure, our method employs trainable patching over features describing a patient, to learn meaningful prototypical parts from structured data. These parts are represented as binary or discretized feature subsets. This allows the model to express prototypes in human-readable terms, enabling alignment with clinical language and case-based reasoning. Our proposed neural network is inherently interpretable and offers interpretable concept-based predictions by comparing the patient's description to learned prototypes in the latent space of the network. In experiments, we demonstrate that the model achieves classification performance competitive to widely used baseline models on medical benchmark datasets, while also offering transparency, bridging the gap between predictive performance and interpretability in clinical decision support.

Ignacy Stępka, Jerzy Stefanowski

Machine learning models in dynamic environments often suffer from concept drift, where changes in the data distribution degrade performance. While detecting this drift is a well-studied topic, explaining how and why the model's decision-making logic changes still remains a significant challenge. In this paper, we introduce a novel methodology to explain concept drift by analyzing the temporal evolution of group-based counterfactual explanations (GCEs). Our approach tracks shifts in the GCEs' cluster centroids and their associated counterfactual action vectors before and after a drift. These evolving GCEs act as an interpretable proxy, revealing structural changes in the model's decision boundary and its underlying rationale. We operationalize this analysis within a three-layer framework that synergistically combines insights from the data layer (distributional shifts), the model layer (prediction disagreement), and our proposed explanation layer. We show that such holistic view allows for a more comprehensive diagnosis of drift, making it possible to distinguish between different root causes, such as a spatial data shift versus a re-labeling of concepts.

Lukasz Sztukiewicz, Ignacy Stępka, Michał Wiliński et al.

The widespread adoption of machine learning systems has raised critical concerns about fairness and bias, making mitigating harmful biases essential for AI development. In this paper, we investigate the relationship between debiasing and removing artifacts in neural networks for computer vision tasks. First, we introduce a set of novel XAI-based metrics that analyze saliency maps to assess shifts in a model's decision-making process. Then, we demonstrate that successful debiasing methods systematically redirect model focus away from protected attributes. Finally, we show that techniques originally developed for artifact removal can be effectively repurposed for improving fairness. These findings provide evidence for the existence of a bidirectional connection between ensuring fairness and removing artifacts corresponding to protected attributes.