Barbara Hammer

Dominik Stallmann, Philip Kenneweg, Barbara Hammer

Transfer learning schemes based on deep networks which have been trained on huge image corpora offer state-of-the-art technologies in computer vision. Here, supervised and semi-supervised approaches constitute efficient technologies which work well with comparably small data sets. Yet, such applications are currently restricted to application domains where suitable deepnetwork models are readily available. In this contribution, we address an important application area in the domain of biotechnology, the automatic analysis of CHO-K1 suspension growth in microfluidic single-cell cultivation, where data characteristics are very dissimilar to existing domains and trained deep networks cannot easily be adapted by classical transfer learning. We propose a novel transfer learning scheme which expands a recently introduced Twin-VAE architecture, which is trained on realistic and synthetic data, and we modify its specialized training procedure to the transfer learning domain. In the specific domain, often only few to no labels exist and annotations are costly. We investigate a novel transfer learning strategy, which incorporates a simultaneous retraining on natural and synthetic data using an invariant shared representation as well as suitable target variables, while it learns to handle unseen data from a different microscopy tech nology. We show the superiority of the variation of our Twin-VAE architecture over the state-of-the-art transfer learning methodology in image processing as well as classical image processing technologies, which persists, even with strongly shortened training times and leads to satisfactory results in this domain. The source code is available at https://github.com/dstallmann/transfer_learning_twinvae, works cross-platform, is open-source and free (MIT licensed) software. We make the data sets available at https://pub.uni-bielefeld.de/record/2960030.

Leona Hennig, Jasmin Brandt, Lukas Fehring et al.

Large-scale hyperparameter optimization (HPO) in automated machine learning (AutoML) consumes substantial computational resources, raising growing concerns about scalability and energy efficiency. Existing methods use prior information heuristically to accelerate both black-box and multi-fidelity settings, but they lack a characterization of how prior informativeness quantitatively reduces sample complexity. In this work, we provide the first distribution-dependent sample complexity bounds for multi-fidelity HPO with priors through the formal lens of fixed-budget best-arm identification. By modeling priors directly over arm means as configuration performance, we derive explicit, distribution-dependent error bounds that quantify the relationship between priors and evaluation budget. Our analysis shows that informative priors, which concentrate probability mass on near-optimal arms, yield reductions in the number of required evaluations, whereas baseline performance is recovered with uninformative or misleading priors. We conduct proof-of-concept experiments on a synthetic benchmark and on LCBench, a common multi-fidelity HPO benchmark for deep learning, to confirm our theoretical results, achieving up to 90% budget reduction while retaining solution quality. Together, our results provide a principled foundation for prior-guided and compute-efficient green AutoML.

Fabian Fumagalli, Maximilian Muschalik, Patrick Kolpaczki et al.

Predominately in explainable artificial intelligence (XAI) research, the Shapley value (SV) is applied to determine feature attributions for any black box model. Shapley interaction indices extend the SV to define any-order feature interactions. Defining a unique Shapley interaction index is an open research question and, so far, three definitions have been proposed, which differ by their choice of axioms. Moreover, each definition requires a specific approximation technique. Here, we propose SHAPley Interaction Quantification (SHAP-IQ), an efficient sampling-based approximator to compute Shapley interactions for arbitrary cardinal interaction indices (CII), i.e. interaction indices that satisfy the linearity, symmetry and dummy axiom. SHAP-IQ is based on a novel representation and, in contrast to existing methods, we provide theoretical guarantees for its approximation quality, as well as estimates for the variance of the point estimates. For the special case of SV, our approach reveals a novel representation of the SV and corresponds to Unbiased KernelSHAP with a greatly simplified calculation. We illustrate the computational efficiency and effectiveness by explaining language, image classification and high-dimensional synthetic models.

Ulrike Kuhl, André Artelt, Barbara Hammer

Counterfactual explanations (CFEs) highlight what changes to a model's input would have changed its prediction in a particular way. CFEs have gained considerable traction as a psychologically grounded solution for explainable artificial intelligence (XAI). Recent innovations introduce the notion of computational plausibility for automatically generated CFEs, enhancing their robustness by exclusively creating plausible explanations. However, practical benefits of such a constraint on user experience and behavior is yet unclear. In this study, we evaluate objective and subjective usability of computationally plausible CFEs in an iterative learning design targeting novice users. We rely on a novel, game-like experimental design, revolving around an abstract scenario. Our results show that novice users actually benefit less from receiving computationally plausible rather than closest CFEs that produce minimal changes leading to the desired outcome. Responses in a post-game survey reveal no differences in terms of subjective user experience between both groups. Following the view of psychological plausibility as comparative similarity, this may be explained by the fact that users in the closest condition experience their CFEs as more psychologically plausible than the computationally plausible counterpart. In sum, our work highlights a little-considered divergence of definitions of computational plausibility and psychological plausibility, critically confirming the need to incorporate human behavior, preferences and mental models already at the design stages of XAI approaches. In the interest of reproducible research, all source code, acquired user data, and evaluation scripts of the current study are available: https://github.com/ukuhl/PlausibleAlienZoo

Fabian Fumagalli, Maximilian Muschalik, Eyke Hüllermeier et al.

Explainable Artificial Intelligence (XAI) has mainly focused on static learning scenarios so far. We are interested in dynamic scenarios where data is sampled progressively, and learning is done in an incremental rather than a batch mode. We seek efficient incremental algorithms for computing feature importance (FI) measures, specifically, an incremental FI measure based on feature marginalization of absent features similar to permutation feature importance (PFI). We propose an efficient, model-agnostic algorithm called iPFI to estimate this measure incrementally and under dynamic modeling conditions including concept drift. We prove theoretical guarantees on the approximation quality in terms of expectation and variance. To validate our theoretical findings and the efficacy of our approaches compared to traditional batch PFI, we conduct multiple experimental studies on benchmark data with and without concept drift.

Vaggelis Dorovatas, Malte Schwerin, Andrew D. Bagdanov et al. · mila

Foundation models have transformed machine learning through large-scale pretraining and increased test-time compute. Despite surpassing human performance in several domains, these models remain fundamentally limited in continuous operation, experience accumulation, and personalization, capabilities that are central to adaptive intelligence. While continual learning research has long targeted these goals, its historical focus on in-weight learning (IWL), i.e., updating a single model's parameters to absorb new knowledge, has rendered catastrophic forgetting a persistent challenge. Our position is that combining the strengths of In-Weight Learning (IWL) and the newly emerged capabilities of In-Context Learning (ICL) through the design of modular memory is the missing piece for continual adaptation at scale. We outline a conceptual framework for modular memory-centric architectures that leverage ICL for rapid adaptation and knowledge accumulation, and IWL for stable updates to model capabilities, charting a practical roadmap toward continually learning agents.

Dennis Zanutto, Christos Michalopoulos, Lydia Tsiami et al.

The highly anticipated 'Battle of the Water Networks' is back with a new challenge for the water community. This competition will be hosted at the 4th International Joint Conference on Water Distribution Systems Analysis and Computing and Control in the Water Industry (WDSA/CCWI 2026), taking place in Paphos, Cyprus, from May 18-21, 2026. This competition embodies the core mission of Water-Futures and the theme for WDSA/CCWI 2026: "Designing the next generation of urban water (and wastewater) systems." The objective is to design and operate a water distribution system over a long-term horizon under deep uncertainty, with interventions applied in stages. For the first time, this challenge features a staged-design approach, unobservable and unknown uncertainties, and incorporates elements of policymaking and artificial intelligence. The solutions will be assessed using a transparent and inspectable open-source evaluation framework.

Maximilian Muschalik, Fabian Fumagalli, Rohit Jagtani et al.

Post-hoc explanation techniques such as the well-established partial dependence plot (PDP), which investigates feature dependencies, are used in explainable artificial intelligence (XAI) to understand black-box machine learning models. While many real-world applications require dynamic models that constantly adapt over time and react to changes in the underlying distribution, XAI, so far, has primarily considered static learning environments, where models are trained in a batch mode and remain unchanged. We thus propose a novel model-agnostic XAI framework called incremental PDP (iPDP) that extends on the PDP to extract time-dependent feature effects in non-stationary learning environments. We formally analyze iPDP and show that it approximates a time-dependent variant of the PDP that properly reacts to real and virtual concept drift. The time-sensitivity of iPDP is controlled by a single smoothing parameter, which directly corresponds to the variance and the approximation error of iPDP in a static learning environment. We illustrate the efficacy of iPDP by showcasing an example application for drift detection and conducting multiple experiments on real-world and synthetic data sets and streams.

Ulrike Kuhl, André Artelt, Barbara Hammer

Counterfactual explanations (CFEs) are a popular approach in explainable artificial intelligence (xAI), highlighting changes to input data necessary for altering a model's output. A CFE can either describe a scenario that is better than the factual state (upward CFE), or a scenario that is worse than the factual state (downward CFE). However, potential benefits and drawbacks of the directionality of CFEs for user behavior in xAI remain unclear. The current user study (N=161) compares the impact of CFE directionality on behavior and experience of participants tasked to extract new knowledge from an automated system based on model predictions and CFEs. Results suggest that upward CFEs provide a significant performance advantage over other forms of counterfactual feedback. Moreover, the study highlights potential benefits of mixed CFEs improving user performance compared to downward CFEs or no explanations. In line with the performance results, users' explicit knowledge of the system is statistically higher after receiving upward CFEs compared to downward comparisons. These findings imply that the alignment between explanation and task at hand, the so-called regulatory fit, may play a crucial role in determining the effectiveness of model explanations, informing future research directions in xAI. To ensure reproducible research, the entire code, underlying models and user data of this study is openly available: https://github.com/ukuhl/DirectionalAlienZoo

André Artelt, Stelios Vrachimis, Demetrios Eliades et al.

Transparency is a major requirement of modern AI based decision making systems deployed in real world. A popular approach for achieving transparency is by means of explanations. A wide variety of different explanations have been proposed for single decision making systems. In practice it is often the case to have a set (i.e. ensemble) of decisions that are used instead of a single decision only, in particular in complex systems. Unfortunately, explanation methods for single decision making systems are not easily applicable to ensembles -- i.e. they would yield an ensemble of individual explanations which are not necessarily consistent, hence less useful and more difficult to understand than a single consistent explanation of all observed phenomena. We propose a novel concept for consistently explaining an ensemble of decisions locally with a single explanation -- we introduce a formal concept, as well as a specific implementation using counterfactual explanations.

Maximilian Muschalik, Fabian Fumagalli, Barbara Hammer et al.

Existing methods for explainable artificial intelligence (XAI), including popular feature importance measures such as SAGE, are mostly restricted to the batch learning scenario. However, machine learning is often applied in dynamic environments, where data arrives continuously and learning must be done in an online manner. Therefore, we propose iSAGE, a time- and memory-efficient incrementalization of SAGE, which is able to react to changes in the model as well as to drift in the data-generating process. We further provide efficient feature removal methods that break (interventional) and retain (observational) feature dependencies. Moreover, we formally analyze our explanation method to show that iSAGE adheres to similar theoretical properties as SAGE. Finally, we evaluate our approach in a thorough experimental analysis based on well-established data sets and data streams with concept drift.

Fabian Hinder, Valerie Vaquet, Johannes Brinkrolf et al.

The notion of concept drift refers to the phenomenon that the distribution generating the observed data changes over time. If drift is present, machine learning models can become inaccurate and need adjustment. While there do exist methods to detect concept drift or to adjust models in the presence of observed drift, the question of explaining drift, i.e., describing the potentially complex and high dimensional change of distribution in a human-understandable fashion, has hardly been considered so far. This problem is of importance since it enables an inspection of the most prominent characteristics of how and where drift manifests itself. Hence, it enables human understanding of the change and it increases acceptance of life-long learning models. In this paper, we present a novel technology characterizing concept drift in terms of the characteristic change of spatial features based on various explanation techniques. To do so, we propose a methodology to reduce the explanation of concept drift to an explanation of models that are trained in a suitable way extracting relevant information regarding the drift. This way a large variety of explanation schemes is available. Thus, a suitable method can be selected for the problem of drift explanation at hand. We outline the potential of this approach and demonstrate its usefulness in several examples.

André Artelt, Kleanthis Malialis, Christos Panayiotou et al.

Concept drift refers to a change in the data distribution affecting the data stream of future samples. Consequently, learning models operating on the data stream might become obsolete, and need costly and difficult adjustments such as retraining or adaptation. Existing methods usually implement a local concept drift adaptation scheme, where either incremental learning of the models is used, or the models are completely retrained when a drift detection mechanism triggers an alarm. This paper proposes an alternative approach in which an unsupervised and model-agnostic concept drift adaptation method at the global level is introduced, based on autoencoders. Specifically, the proposed method aims to ``unlearn'' the concept drift without having to retrain or adapt any of the learning models operating on the data. An extensive experimental evaluation is conducted in two application domains. We consider a realistic water distribution network with more than 30 models in-place, from which we create 200 simulated data sets / scenarios. We further consider an image-related task to demonstrate the effectiveness of our method.

Ulrike Kuhl, André Artelt, Barbara Hammer

To foster usefulness and accountability of machine learning (ML), it is essential to explain a model's decisions in addition to evaluating its performance. Accordingly, the field of explainable artificial intelligence (XAI) has resurfaced as a topic of active research, offering approaches to address the "how" and "why" of automated decision-making. Within this domain, counterfactual explanations (CFEs) have gained considerable traction as a psychologically grounded approach to generate post-hoc explanations. To do so, CFEs highlight what changes to a model's input would have changed its prediction in a particular way. However, despite the introduction of numerous CFE approaches, their usability has yet to be thoroughly validated at the human level. Thus, to advance the field of XAI, we introduce the Alien Zoo, an engaging, web-based and game-inspired experimental framework. The Alien Zoo provides the means to evaluate usability of CFEs for gaining new knowledge from an automated system, targeting novice users in a domain-general context. As a proof of concept, we demonstrate the practical efficacy and feasibility of this approach in a user study. Our results suggest that users benefit from receiving CFEs compared to no explanation, both in terms of objective performance in the proposed iterative learning task, and subjective usability. With this work, we aim to equip research groups and practitioners with the means to easily run controlled and well-powered user studies to complement their otherwise often more technology-oriented work. Thus, in the interest of reproducible research, we provide the entire code, together with the underlying models and user data.

Inaam Ashraf, Luca Hermes, André Artelt et al.

We investigate the task of missing value estimation in graphs as given by water distribution systems (WDS) based on sparse signals as a representative machine learning challenge in the domain of critical infrastructure. The underlying graphs have a comparably low node degree and high diameter, while information in the graph is globally relevant, hence graph neural networks face the challenge of long-term dependencies. We propose a specific architecture based on message passing which displays excellent results for a number of benchmark tasks in the WDS domain. Further, we investigate a multi-hop variation, which requires considerably less resources and opens an avenue towards big WDS graphs.

Andrea Castellani, Sebastian Schmitt, Barbara Hammer

Data stream classification is an important problem in the field of machine learning. Due to the non-stationary nature of the data where the underlying distribution changes over time (concept drift), the model needs to continuously adapt to new data statistics. Stream-based Active Learning (AL) approaches address this problem by interactively querying a human expert to provide new data labels for the most recent samples, within a limited budget. Existing AL strategies assume that labels are immediately available, while in a real-world scenario the expert requires time to provide a queried label (verification latency), and by the time the requested labels arrive they may not be relevant anymore. In this article, we investigate the influence of finite, time-variable, and unknown verification delay, in the presence of concept drift on AL approaches. We propose PRopagate (PR), a latency independent utility estimator which also predicts the requested, but not yet known, labels. Furthermore, we propose a drift-dependent dynamic budget strategy, which uses a variable distribution of the labelling budget over time, after a detected drift. Thorough experimental evaluation, with both synthetic and real-world non-stationary datasets, and different settings of verification latency and budget are conducted and analyzed. We empirically show that the proposed method consistently outperforms the state-of-the-art. Additionally, we demonstrate that with variable budget allocation in time, it is possible to boost the performance of AL strategies, without increasing the overall labeling budget.

André Artelt, Barbara Hammer

Machine learning based decision making systems applied in safety critical areas require reliable high certainty predictions. For this purpose, the system can be extended by an reject option which allows the system to reject inputs where only a prediction with an unacceptably low certainty would be possible. While being able to reject uncertain samples is important, it is also of importance to be able to explain why a particular sample was rejected. With the ongoing rise of eXplainable AI (XAI), a lot of explanation methodologies for machine learning based systems have been developed -- explaining reject options, however, is still a novel field where only very little prior work exists. In this work, we propose to explain rejects by semifactual explanations, an instance of example-based explanation methods, which them self have not been widely considered in the XAI community yet. We propose a conceptual modeling of semifactual explanations for arbitrary reject options and empirically evaluate a specific implementation on a conformal prediction based reject option.

Fabian Hinder, Valerie Vaquet, Barbara Hammer

The world surrounding us is subject to constant change. These changes, frequently described as concept drift, influence many industrial and technical processes. As they can lead to malfunctions and other anomalous behavior, which may be safety-critical in many scenarios, detecting and analyzing concept drift is crucial. In this paper, we provide a literature review focusing on concept drift in unsupervised data streams. While many surveys focus on supervised data streams, so far, there is no work reviewing the unsupervised setting. However, this setting is of particular relevance for monitoring and anomaly detection which are directly applicable to many tasks and challenges in engineering. This survey provides a taxonomy of existing work on drift detection. Besides, it covers the current state of research on drift localization in a systematic way. In addition to providing a systematic literature review, this work provides precise mathematical definitions of the considered problems and contains standardized experiments on parametric artificial datasets allowing for a direct comparison of different strategies for detection and localization. Thereby, the suitability of different schemes can be analyzed systematically and guidelines for their usage in real-world scenarios can be provided. Finally, there is a section on the emerging topic of explaining concept drift.

André Artelt, Barbara Hammer

Counterfactual explanations are a popular type of explanation for making the outcomes of a decision making system transparent to the user. Counterfactual explanations tell the user what to do in order to change the outcome of the system in a desirable way. However, it was recently discovered that the recommendations of what to do can differ significantly in their complexity between protected groups of individuals. Providing more difficult recommendations of actions to one group leads to a disadvantage of this group compared to other groups. In this work we propose a model-agnostic method for computing counterfactual explanations that do not differ significantly in their complexity between protected groups.

André Artelt, Roel Visser, Barbara Hammer

The application of machine learning based decision making systems in safety critical areas requires reliable high certainty predictions. Reject options are a common way of ensuring a sufficiently high certainty of predictions made by the system. While being able to reject uncertain samples is important, it is also of importance to be able to explain why a particular sample was rejected. However, explaining general reject options is still an open problem. We propose a model agnostic method for locally explaining arbitrary reject options by means of interpretable models and counterfactual explanations.

Riza Velioglu, Jan Philip Göpfert, André Artelt et al.

In modern business processes, the amount of data collected has increased substantially in recent years. Because this data can potentially yield valuable insights, automated knowledge extraction based on process mining has been proposed, among other techniques, to provide users with intuitive access to the information contained therein. At present, the majority of technologies aim to reconstruct explicit business process models. These are directly interpretable but limited concerning the integration of diverse and real-valued information sources. On the other hand, Machine Learning (ML) benefits from the vast amount of data available and can deal with high-dimensional sources, yet it has rarely been applied to being used in processes. In this contribution, we evaluate the capability of modern Transformer architectures as well as more classical ML technologies of modeling process regularities, as can be quantitatively evaluated by their prediction capability. In addition, we demonstrate the capability of attentional properties and feature relevance determination by highlighting features that are crucial to the processes' predictive abilities. We demonstrate the efficacy of our approach using five benchmark datasets and show that the ML models are capable of predicting critical outcomes and that the attention mechanisms or XAI components offer new insights into the underlying processes.

Valerie Vaquet, Fabian Hinder, Jonas Vaquet et al.

Facing climate change, the already limited availability of drinking water will decrease in the future rendering drinking water an increasingly scarce resource. Considerable amounts of it are lost through leakages in water transportation and distribution networks. Thus, anomaly detection and localization, in particular for leakages, are crucial but challenging tasks due to the complex interactions and changing demands in water distribution networks. In this work, we analyze the effects of anomalies on the dynamics of critical infrastructure systems by modeling the networks employing Bayesian networks. We then discuss how the problem is connected to and can be considered through the lens of concept drift. In particular, we argue that model-based explanations of concept drift are a promising tool for localizing anomalies given limited information about the network. The methodology is experimentally evaluated using realistic benchmark scenarios. To showcase that our methodology applies to critical infrastructure more generally, in addition to considering leakages and sensor faults in water systems, we showcase the suitability of the derived technique to localize sensor faults in power systems.

Janine Strotherm, Alissa Müller, Barbara Hammer et al.

The more AI-assisted decisions affect people's lives, the more important the fairness of such decisions becomes. In this chapter, we provide an introduction to research on fairness in machine learning. We explain the main fairness definitions and strategies for achieving fairness using concrete examples and place fairness research in the European context. Our contribution is aimed at an interdisciplinary audience and therefore avoids mathematical formulation but emphasizes visualizations and examples. -- Je mehr KI-gestützte Entscheidungen das Leben von Menschen betreffen, desto wichtiger ist die Fairness solcher Entscheidungen. In diesem Kapitel geben wir eine Einführung in die Forschung zu Fairness im maschinellen Lernen. Wir erklären die wesentlichen Fairness-Definitionen und Strategien zur Erreichung von Fairness anhand konkreter Beispiele und ordnen die Fairness-Forschung in den europäischen Kontext ein. Unser Beitrag richtet sich dabei an ein interdisziplinäres Publikum und verzichtet daher auf die mathematische Formulierung sondern betont Visualisierungen und Beispiele.

Jonathan Jakob, André Artelt, Martina Hasenjäger et al.

Water distribution networks are a key component of modern infrastructure for housing and industry. They transport and distribute water via widely branched networks from sources to consumers. In order to guarantee a working network at all times, the water supply company continuously monitors the network and takes actions when necessary -- e.g. reacting to leakages, sensor faults and drops in water quality. Since real world networks are too large and complex to be monitored by a human, algorithmic monitoring systems have been developed. A popular type of such systems are residual based anomaly detection systems that can detect events such as leakages and sensor faults. For a continuous high quality monitoring, it is necessary for these systems to adapt to changed demands and presence of various anomalies. In this work, we propose an adaption of the incremental SAM-kNN classifier for regression to build a residual based anomaly detection system for water distribution networks that is able to adapt to any kind of change.

Sarah Schröder, Alexander Schulz, Barbara Hammer

With the enourmous popularity of large language models, many researchers have raised ethical concerns regarding social biases incorporated in such models. Several methods to measure social bias have been introduced, but apparently these methods do not necessarily agree regarding the presence or severity of bias. Furthermore, some works have shown theoretical issues or severe limitations with certain bias measures. For that reason, we introduce SAME, a novel bias score for semantic bias in embeddings. We conduct a thorough theoretical analysis as well as experiments to show its benefits compared to similar bias scores from the literature. We further highlight a substantial relation of semantic bias measured by SAME with downstream bias, a connection that has recently been argued to be negligible. Instead, we show that SAME is capable of measuring semantic bias and identify potential causes for social bias in downstream tasks.

Andrea Cossu, Federico Giannini, Giacomo Ziffer et al.

Continual Learning (CL) and Streaming Machine Learning (SML) study the ability of agents to learn from a stream of non-stationary data. Despite sharing some similarities, they address different and complementary challenges. While SML focuses on rapid adaptation after changes (concept drifts), CL aims to retain past knowledge when learning new tasks. After a brief introduction to CL and SML, we discuss Streaming Continual Learning (SCL), an emerging paradigm providing a unifying solution to real-world problems, which may require both SML and CL abilities. We claim that SCL can i) connect the CL and SML communities, motivating their work towards the same goal, and ii) foster the design of hybrid approaches that can quickly adapt to new information (as in SML) without forgetting previous knowledge (as in CL). We conclude the paper with a motivating example and a set of experiments, highlighting the need for SCL by showing how CL and SML alone struggle in achieving rapid adaptation and knowledge retention.

Fabian Hinder, Valerie Vaquet, Johannes Brinkrolf et al.

Concept drift -- the change of the distribution over time -- poses significant challenges for learning systems and is of central interest for monitoring. Understanding drift is thus paramount, and drift localization -- determining which samples are affected by the drift -- is essential. While several approaches exist, most rely on local testing schemes, which tend to fail in high-dimensional, low-signal settings. In this work, we consider a fundamentally different approach based on conformal predictions. We discuss and show the shortcomings of common approaches and demonstrate the performance of our approach on state-of-the-art image datasets.

Valerie Vaquet, Fabian Hinder, Johannes Brinkrolf et al.

Learning from non-stationary data streams is a research direction that gains increasing interest as more data in form of streams becomes available, for example from social media, smartphones, or industrial process monitoring. Most approaches assume that the ground truth of the samples becomes available (possibly with some delay) and perform supervised online learning in the test-then-train scheme. While this assumption might be valid in some scenarios, it does not apply to all settings. In this work, we focus on scarcely labeled data streams and explore the potential of self-labeling in gradually drifting data streams. We formalize this setup and propose a novel online $k$-nn classifier that combines self-labeling and demand-based active learning.

Fabian Hinder, Valerie Vaquet, Johannes Brinkrolf et al.

The notion of concept drift refers to the phenomenon that the distribution generating the observed data changes over time. If drift is present, machine learning models may become inaccurate and need adjustment. Many technologies for learning with drift rely on the interleaved test-train error (ITTE) as a quantity which approximates the model generalization error and triggers drift detection and model updates. In this work, we investigate in how far this procedure is mathematically justified. More precisely, we relate a change of the ITTE to the presence of real drift, i.e., a changed posterior, and to a change of the training result under the assumption of optimality. We support our theoretical findings by empirical evidence for several learning algorithms, models, and datasets.

Fabian Hinder, André Artelt, Valerie Vaquet et al.

The notion of concept drift refers to the phenomenon that the data generating distribution changes over time; as a consequence machine learning models may become inaccurate and need adjustment. In this paper we consider the problem of detecting those change points in unsupervised learning. Many unsupervised approaches rely on the discrepancy between the sample distributions of two time windows. This procedure is noisy for small windows, hence prone to induce false positives and not able to deal with more than one drift event in a window. In this paper we rely on structural properties of drift induced signals, which use spectral properties of kernel embedding of distributions. Based thereon we derive a new unsupervised drift detection algorithm, investigate its mathematical properties, and demonstrate its usefulness in several experiments.

Tristan Kenneweg, Philip Kenneweg, Barbara Hammer

Retrieval Augmented Generation (RAG) systems have seen huge popularity in augmenting Large-Language Model (LLM) outputs with domain specific and time sensitive data. Very recently a shift is happening from simple RAG setups that query a vector database for additional information with every user input to more sophisticated forms of RAG. However, different concrete approaches compete on mostly anecdotal evidence at the moment. In this paper we present a rigorous dataset creation and evaluation workflow to quantitatively compare different RAG strategies. We use a dataset created this way for the development and evaluation of a boolean agent RAG setup: A system in which a LLM can decide whether to query a vector database or not, thus saving tokens on questions that can be answered with internal knowledge. We publish our code and generated dataset online.

Felix Störck, Fabian Hinder, Barbara Hammer

With the on-going integration of machine learning systems into the everyday social life of millions the notion of fairness becomes an ever increasing priority in their development. Fairness notions commonly rely on protected attributes to assess potential biases. Here, the majority of literature focuses on discrete setups regarding both target and protected attributes. The literature on continuous attributes especially in conjunction with regression -- we refer to this as \emph{continuous fairness} -- is scarce. A common strategy is iterative null-space projection which as of now has only been explored for linear models or embeddings such as obtained by a non-linear encoder. We improve on this by generalizing to kernel methods, significantly extending the scope. This yields a model and fairness-score agnostic method for kernel embeddings applicable to continuous protected attributes. We demonstrate that our novel approach in conjunction with Support Vector Regression (SVR) provides competitive or improved performance across multiple datasets in comparisons to other contemporary methods.

Philip Kenneweg, Tristan Kenneweg, Fabian Fumagalli et al.

In recent studies, line search methods have been demonstrated to significantly enhance the performance of conventional stochastic gradient descent techniques across various datasets and architectures, while making an otherwise critical choice of learning rate schedule superfluous. In this paper, we identify problems of current state-of-the-art of line search methods, propose enhancements, and rigorously assess their effectiveness. Furthermore, we evaluate these methods on orders of magnitude larger datasets and more complex data domains than previously done. More specifically, we enhance the Armijo line search method by speeding up its computation and incorporating a momentum term into the Armijo criterion, making it better suited for stochastic mini-batching. Our optimization approach outperforms both the previous Armijo implementation and a tuned learning rate schedule for the Adam and SGD optimizers. Our evaluation covers a diverse range of architectures, such as Transformers, CNNs, and MLPs, as well as data domains, including NLP and image data. Our work is publicly available as a Python package, which provides a simple Pytorch optimizer.

Aleksei Liuliakov, Luca Hermes, Barbara Hammer

Reinforcement learning fine-tuning has proven effective for steering generative diffusion models toward desired properties in image and molecular domains. Graph diffusion models have similarly been applied to combinatorial structure generation, including neural architecture search (NAS). However, neural architectures are directed acyclic graphs (DAGs) where edge direction encodes functional semantics such as data flow-information that existing graph diffusion methods, designed for undirected structures, discard. We propose Directed Graph Policy Optimization (DGPO), which extends reinforcement learning fine-tuning of discrete graph diffusion models to DAGs via topological node ordering and positional encoding. Validated on NAS-Bench-101 and NAS-Bench-201, DGPO matches the benchmark optimum on all three NAS-Bench-201 tasks (91.61%, 73.49%, 46.77%). The central finding is that the model learns transferable structural priors: pretrained on only 7% of the search space, it generates near-oracle architectures after fine-tuning, within 0.32 percentage points of the full-data model and extrapolating 7.3 percentage points beyond its training ceiling. Bidirectional control experiments confirm genuine reward-driven steering, with inverse optimization reaching near random-chance accuracy (9.5%). These results demonstrate that reinforcement learning-steered discrete diffusion, once extended to handle directionality, provides a controllable generative framework for directed combinatorial structures.

Michiel Straat, Thorben Markmann, Sebastian Peitz et al.

Chaotic convective flows arise in many real-world systems, such as microfluidic devices and chemical reactors. Stabilizing these flows is highly desirable but remains challenging, particularly in chaotic regimes where conventional control methods often fail. Reinforcement Learning (RL) has shown promise for control in laminar flow settings, but its ability to generalize and remain robust under chaotic and turbulent dynamics is not well explored, despite being critical for real-world deployment. In this work, we improve the practical feasibility of RL-based control of such flows focusing on Rayleigh-Bénard Convection (RBC), a canonical model for convective heat transport. To enhance generalization and sample efficiency, we introduce domain-informed RL agents that are trained using Proximal Policy Optimization across diverse initial conditions and flow regimes. We incorporate domain knowledge in the reward function via a term that encourages Bénard cell merging, as an example of a desirable macroscopic property. In laminar flow regimes, the domain-informed RL agents reduce convective heat transport by up to 33%, and in chaotic flow regimes, they still achieve a 10% reduction, which is significantly better than the conventional controllers used in practice. We compare the domain-informed to uninformed agents: Our results show that the domain-informed reward design results in steady flows, faster convergence during training, and generalization across flow regimes without retraining. Our work demonstrates that elegant domain-informed priors can greatly enhance the robustness of RL-based control of chaotic flows, bringing real-world deployment closer.

Christian Internò, Andrea Castellani, Sebastian Schmitt et al.

Industrial Non-Intrusive Load Monitoring (NILM) is limited by the scarcity of high-quality datasets and the complex variability of industrial energy consumption patterns. To address data scarcity and privacy issues, we introduce the Synthetic Industrial Dataset for Energy Disaggregation (SIDED), an open-source dataset generated using Digital Twin simulations. SIDED includes three types of industrial facilities across three different geographic locations, capturing diverse appliance behaviors, weather conditions, and load profiles. We also propose the Appliance-Modulated Data Augmentation (AMDA) method, a computationally efficient technique that enhances NILM model generalization by intelligently scaling appliance power contributions based on their relative impact. We show in experiments that NILM models trained with AMDA-augmented data significantly improve the disaggregation of energy consumption of complex industrial appliances like combined heat and power systems. Specifically, in our out-of-sample scenarios, models trained with AMDA achieved a Normalized Disaggregation Error of 0.093, outperforming models trained without data augmentation (0.451) and those trained with random data augmentation (0.290). Data distribution analyses confirm that AMDA effectively aligns training and test data distributions, enhancing model generalization.

Alexander Schulz, Fabian Hinder, Barbara Hammer

Machine learning algorithms using deep architectures have been able to implement increasingly powerful and successful models. However, they also become increasingly more complex, more difficult to comprehend and easier to fool. So far, most methods in the literature investigate the decision of the model for a single given input datum. In this paper, we propose to visualize a part of the decision function of a deep neural network together with a part of the data set in two dimensions with discriminative dimensionality reduction. This enables us to inspect how different properties of the data are treated by the model, such as outliers, adversaries or poisoned data. Further, the presented approach is complementary to the mentioned interpretation methods from the literature and hence might be even more useful in combination with those. Code is available at https://github.com/LucaHermes/DeepView .

Lukas Pfannschmidt, Christina Göpfert, Ursula Neumann et al.

Most existing feature selection methods are insufficient for analytic purposes as soon as high dimensional data or redundant sensor signals are dealt with since features can be selected due to spurious effects or correlations rather than causal effects. To support the finding of causal features in biomedical experiments, we hereby present FRI, an open source Python library that can be used to identify all-relevant variables in linear classification and (ordinal) regression problems. Using the recently proposed feature relevance method, FRI is able to provide the base for further general experimentation or in specific can facilitate the search for alternative biomarkers. It can be used in an interactive context, by providing model manipulation and visualization methods, or in a batch process as a filter method.

Andrea Moleri, Christian Internò, Ali Raza et al.

Federated Learning (FL) enables distributed optimization without compromising data sovereignty. Yet, where local label distributions are mutually exclusive, standard weight aggregation fails due to conflicting optimization trajectories. Often, FL methods rely on pretrained foundation models, introducing unrealistic assumptions. We introduce FederatedFactory, a zero-dependency framework that inverts the unit of federation from discriminative parameters to generative priors. By exchanging generative modules in a single communication round, our architecture supports ex nihilo synthesis of universally class balanced datasets, eliminating gradient conflict and external prior bias entirely. Evaluations across diverse medical imagery benchmarks, including MedMNIST and ISIC2019, demonstrate that our approach recovers centralized upper-bound performance. Under pathological heterogeneity, it lifts baseline accuracy from a collapsed 11.36% to 90.57% on CIFAR-10 and restores ISIC2019 AUROC to 90.57%. Additionally, this framework facilitates exact modular unlearning through the deterministic deletion of specific generative modules.

Christian Internò, Jumpei Yamaguchi, Loren Amdahl-Culleton et al.

Determining whether neural models internalize physical laws as world models, rather than exploiting statistical shortcuts, remains challenging, especially under out-of-distribution (OOD) shifts. Standard evaluations often test latent capability via downstream adaptation (e.g., fine-tuning or high-capacity probes), but such interventions can change the representations being measured and thus confound what was learned during self-supervised learning (SSL). We propose a non-invasive evaluation protocol, PhyIP. We test whether physical quantities are linearly decodable from frozen representations, motivated by the linear representation hypothesis. Across fluid dynamics and orbital mechanics, we find that when SSL achieves low error, latent structure becomes linearly accessible. PhyIP recovers internal energy and Newtonian inverse-square scaling on OOD tests (e.g., $ρ> 0.90$). In contrast, adaptation-based evaluations can collapse this structure ($ρ\approx 0.05$). These findings suggest that adaptation-based evaluation can obscure latent structures and that low-capacity probes offer a more accurate evaluation of physical world models.

Maximilian Muschalik, Fabian Fumagalli, Barbara Hammer et al.

While shallow decision trees may be interpretable, larger ensemble models like gradient-boosted trees, which often set the state of the art in machine learning problems involving tabular data, still remain black box models. As a remedy, the Shapley value (SV) is a well-known concept in explainable artificial intelligence (XAI) research for quantifying additive feature attributions of predictions. The model-specific TreeSHAP methodology solves the exponential complexity for retrieving exact SVs from tree-based models. Expanding beyond individual feature attribution, Shapley interactions reveal the impact of intricate feature interactions of any order. In this work, we present TreeSHAP-IQ, an efficient method to compute any-order additive Shapley interactions for predictions of tree-based models. TreeSHAP-IQ is supported by a mathematical framework that exploits polynomial arithmetic to compute the interaction scores in a single recursive traversal of the tree, akin to Linear TreeSHAP. We apply TreeSHAP-IQ on state-of-the-art tree ensembles and explore interactions on well-established benchmark datasets.

Fabian Fumagalli, Maximilian Muschalik, Patrick Kolpaczki et al.

The Shapley value (SV) is a prevalent approach of allocating credit to machine learning (ML) entities to understand black box ML models. Enriching such interpretations with higher-order interactions is inevitable for complex systems, where the Shapley Interaction Index (SII) is a direct axiomatic extension of the SV. While it is well-known that the SV yields an optimal approximation of any game via a weighted least square (WLS) objective, an extension of this result to SII has been a long-standing open problem, which even led to the proposal of an alternative index. In this work, we characterize higher-order SII as a solution to a WLS problem, which constructs an optimal approximation via SII and $k$-Shapley values ($k$-SII). We prove this representation for the SV and pairwise SII and give empirically validated conjectures for higher orders. As a result, we propose KernelSHAP-IQ, a direct extension of KernelSHAP for SII, and demonstrate state-of-the-art performance for feature interactions.

Clarissa Sabrina Arlinghaus, Tristan Kenneweg, Barbara Hammer et al.

Large language models (LLMs) such as ChatGPT are increasingly integrated into high-stakes decision-making, yet little is known about their susceptibility to social influence. We conducted three preregistered conformity experiments with GPT-4o in a hiring context. In a baseline study, GPT consistently favored the same candidate (Profile C), reported moderate expertise (M = 3.01) and high certainty (M = 3.89), and rarely changed its choice. In Study 1 (GPT + 8), GPT faced unanimous opposition from eight simulated partners and almost always conformed (99.9%), reporting lower certainty and significantly elevated self-reported informational and normative conformity (p < .001). In Study 2 (GPT + 1), GPT interacted with a single partner and still conformed in 40.2% of disagreement trials, reporting less certainty and more normative conformity. Across studies, results demonstrate that GPT does not act as an independent observer but adapts to perceived social consensus. These findings highlight risks of treating LLMs as neutral decision aids and underline the need to elicit AI judgments prior to exposing them to human opinions.

Inaam Ashraf, Janine Strotherm, Luca Hermes et al.

Water distribution systems (WDS) are an integral part of critical infrastructure which is pivotal to urban development. As 70% of the world's population will likely live in urban environments in 2050, efficient simulation and planning tools for WDS play a crucial role in reaching UN's sustainable developmental goal (SDG) 6 - "Clean water and sanitation for all". In this realm, we propose a novel and efficient machine learning emulator, more precisely, a physics-informed deep learning (DL) model, for hydraulic state estimation in WDS. Using a recursive approach, our model only needs a few graph convolutional neural network (GCN) layers and employs an innovative algorithm based on message passing. Unlike conventional machine learning tasks, the model uses hydraulic principles to infer two additional hydraulic state features in the process of reconstructing the available ground truth feature in an unsupervised manner. To the best of our knowledge, this is the first DL approach to emulate the popular hydraulic simulator EPANET, utilizing no additional information. Like most DL models and unlike the hydraulic simulator, our model demonstrates vastly faster emulation times that do not increase drastically with the size of the WDS. Moreover, we achieve high accuracy on the ground truth and very similar results compared to the hydraulic simulator as demonstrated through experiments on five real-world WDS datasets.

Fabian Fumagalli, Maximilian Muschalik, Eyke Hüllermeier et al.

Feature-based explanations, using perturbations or gradients, are a prevalent tool to understand decisions of black box machine learning models. Yet, differences between these methods still remain mostly unknown, which limits their applicability for practitioners. In this work, we introduce a unified framework for local and global feature-based explanations using two well-established concepts: functional ANOVA (fANOVA) from statistics, and the notion of value and interaction from cooperative game theory. We introduce three fANOVA decompositions that determine the influence of feature distributions, and use game-theoretic measures, such as the Shapley value and interactions, to specify the influence of higher-order interactions. Our framework combines these two dimensions to uncover similarities and differences between a wide range of explanation techniques for features and groups of features. We then empirically showcase the usefulness of our framework on synthetic and real-world datasets.

Riza Velioglu, Petra Bevandic, Robin Chan et al.

This paper introduces Virtual Try-Off (VTOFF), a novel task generating standardized garment images from single photos of clothed individuals. Unlike Virtual Try-On (VTON), which digitally dresses models, VTOFF extracts canonical garment images, demanding precise reconstruction of shape, texture, and complex patterns, enabling robust evaluation of generative model fidelity. We propose TryOffDiff, adapting Stable Diffusion with SigLIP-based visual conditioning to deliver high-fidelity reconstructions. Experiments on VITON-HD and Dress Code datasets show that TryOffDiff outperforms adapted pose transfer and VTON baselines. We observe that traditional metrics such as SSIM inadequately reflect reconstruction quality, prompting our use of DISTS for reliable assessment. Our findings highlight VTOFF's potential to improve e-commerce product imagery, advance generative model evaluation, and guide future research on high-fidelity reconstruction. Demo, code, and models are available at: https://rizavelioglu.github.io/tryoffdiff

Thorben Markmann, Michiel Straat, Barbara Hammer

Several related works have introduced Koopman-based Machine Learning architectures as a surrogate model for dynamical systems. These architectures aim to learn non-linear measurements (also known as observables) of the system's state that evolve by a linear operator and are, therefore, amenable to model-based linear control techniques. So far, mainly simple systems have been targeted, and Koopman architectures as reduced-order models for more complex dynamics have not been fully explored. Hence, we use a Koopman-inspired architecture called the Linear Recurrent Autoencoder Network (LRAN) for learning reduced-order dynamics in convection flows of a Rayleigh Bénard Convection (RBC) system at different amounts of turbulence. The data is obtained from direct numerical simulations of the RBC system. A traditional fluid dynamics method, the Kernel Dynamic Mode Decomposition (KDMD), is used to compare the LRAN. For both methods, we performed hyperparameter sweeps to identify optimal settings. We used a Normalized Sum of Square Error measure for the quantitative evaluation of the models, and we also studied the model predictions qualitatively. We obtained more accurate predictions with the LRAN than with KDMD in the most turbulent setting. We conjecture that this is due to the LRAN's flexibility in learning complicated observables from data, thereby serving as a viable surrogate model for the main structure of fluid dynamics in turbulent convection settings. In contrast, KDMD was more effective in lower turbulence settings due to the repetitiveness of the convection flow. The feasibility of Koopman-based surrogate models for turbulent fluid flows opens possibilities for efficient model-based control techniques useful in a variety of industrial settings.

Valerie Vaquet, Fabian Hinder, Barbara Hammer

Leakages are a major risk in water distribution networks as they cause water loss and increase contamination risks. Leakage detection is a difficult task due to the complex dynamics of water distribution networks. In particular, small leakages are hard to detect. From a machine-learning perspective, leakages can be modeled as concept drift. Thus, a wide variety of drift detection schemes seems to be a suitable choice for detecting leakages. In this work, we explore the potential of model-loss-based and distribution-based drift detection methods to tackle leakage detection. We additionally discuss the issue of temporal dependencies in the data and propose a way to cope with it when applying distribution-based detection. We evaluate different methods systematically for leakages of different sizes and detection times. Additionally, we propose a first drift-detection-based technique for localizing leakages.

Aleksei Liuliakov, Alexander Schulz, Luca Hermes et al.

With the growing digitalization all over the globe, the relevance of network security becomes increasingly important. Machine learning-based intrusion detection constitutes a promising approach for improving security, but it bears several challenges. These include the requirement to detect novel and unseen network events, as well as specific data properties, such as events over time together with the inherent graph structure of network communication. In this work, we propose a novel intrusion detection method, TGN-SVDD, which builds upon modern dynamic graph modelling and deep anomaly detection. We demonstrate its superiority over several baselines for realistic intrusion detection data and suggest a more challenging variant of the latter.

Christian Internò, Markus Olhofer, Yaochu Jin et al.

Federated learning (FL) has emerged as a groundbreaking paradigm in machine learning (ML), offering privacy-preserving collaborative model training across diverse datasets. Despite its promise, FL faces significant hurdles in non-identically and independently distributed (non-IID) data scenarios, where most existing methods often struggle with data heterogeneity and lack robustness in performance. This paper introduces Federated Loss Exploration (FedLEx), an innovative approach specifically designed to tackle these challenges. FedLEx distinctively addresses the shortcomings of existing FL methods in non-IID settings by optimizing its learning behavior for scenarios in which assumptions about data heterogeneity are impractical or unknown. It employs a federated loss exploration technique, where clients contribute to a global guidance matrix by calculating gradient deviations for model parameters. This matrix serves as a strategic compass to guide clients' gradient updates in subsequent FL rounds, thereby fostering optimal parameter updates for the global model. FedLEx effectively navigates the complex loss surfaces inherent in non-IID data, enhancing knowledge transfer in an efficient manner, since only a small number of epochs and small amount of data are required to build a strong global guidance matrix that can achieve model convergence without the need for additional data sharing or data distribution statics in a large client scenario. Our extensive experiments with state-of-the art FL algorithms demonstrate significant improvements in performance, particularly under realistic non-IID conditions, thus highlighting FedLEx's potential to overcome critical barriers in diverse FL applications.