Textual data used to train large language models (LLMs) exhibits multifaceted bias manifestations, encompassing harmful language and skewed demographic distributions. Regulations such as the EU AI Act require identifying and mitigating biases against protected groups in data, with the ultimate goal of preventing unfair model outputs. However, practical guidance and operationalization are lacking. We propose a comprehensive data bias detection and mitigation pipeline comprising four components that address two data bias types, namely representation bias and (explicit) stereotypes for a configurable sensitive attribute. First, we leverage LLM-generated word lists, created according to defined quality criteria, to detect relevant group labels. Second, representation bias is quantified using the Demographic Representation Score. Third, we detect and mitigate stereotypes using sociolinguistically informed filtering. Finally, we mitigate representation bias through Grammar- and Context-Aware Counterfactual Data Augmentation. We conduct a twofold evaluation using gender, religion , and age as examples. First, we evaluate the effectiveness of each individual component on data debiasing through human validation and baseline comparison. The findings demonstrate that we successfully reduce representation bias and (explicit) stereotypes in a text dataset. Second, we evaluate the effect of data debiasing on model bias by benchmarking several models (0.6B-8B parameters) fine-tuned on the debiased text dataset. This evaluation reveals that LLMs fine-tuned on debiased data do not consistently show improved performance on bias benchmarks. These results expose critical gaps in current evaluation methodologies and highlight the need for targeted data interventions to address manifested model bias.

Abstract Graph neural networks (GNNs) are among the most widely used methods for node classification in graphs. A common strategy to improve their predictive performance is to enrich nodes with additional features. A weakness of this method is that the set of appropriate features can vary from graph to graph. We address this shortcoming by proposing a novel method. In a preprocessing step, a first GNN is trained on a set of graphs with varying structural properties, using a candidate set of node features fixed in advance. The resulting GNN model is then used to predict the most relevant features from the candidate set for unseen target graphs, which are later processed for node classification. For each target graph, a second GNN is trained on the graph, which is enriched with the node feature vectors calculated for the features selected by the first GNN. A key advantage of the proposed method is that the features are selected without computing the candidate features for the target graph. Our experimental results on synthetic and real-world graphs show that even a few features selected in this way is sufficient to significantly improve the predictive performance of GNNs that use either none or all of the candidate features. Moreover, the time needed to learn the second GNN for the target graph can be reduced by up to two orders of magnitude.

Abstract Despite of its importance for safe machine learning, uncertainty quantification for neural networks is far from being solved. State-of-the-art approaches to estimate neural uncertainties are often hybrid, combining parametric models with explicit or implicit (dropout-based) ensembling. We take another pathway and propose a novel approach to uncertainty quantification for regression tasks, Wasserstein dropout, that is purely non-parametric. Technically, it captures aleatoric uncertainty by means of dropout-based sub-network distributions. This is accomplished by a new objective which minimizes the Wasserstein distance between the label distribution and the model distribution. An extensive empirical analysis shows that Wasserstein dropout outperforms state-of-the-art methods, on vanilla test data as well as under distributional shift in terms of producing more accurate and stable uncertainty estimates.

The majority of popular graph kernels is based on the concept of Haussler's R-convolution kernel and defines graph similarities in terms of mutual substructures. In this work, we enrich these similarity measures by considering graph filtrations: Using meaningful orders on the set of edges, which allow to construct a sequence of nested graphs, we can consider a graph at multiple granularities. A key concept of our approach is to track graph features over the course of such graph resolutions. Rather than to simply compare frequencies of features in graphs, this allows for their comparison in terms of when and for how long they exist in the sequences. In this work, we propose a family of graph kernels that incorporate these existence intervals of features. While our approach can be applied to arbitrary graph features, we particularly highlight Weisfeiler-Lehman vertex labels, leading to efficient kernels. We show that using Weisfeiler-Lehman labels over certain filtrations strictly increases the expressive power over the ordinary Weisfeiler-Lehman procedure in terms of deciding graph isomorphism. In fact, this result directly yields more powerful graph kernels based on such features and has implications to graph neural networks due to their close relationship to the Weisfeiler-Lehman method. We empirically validate the expressive power of our graph kernels and show significant improvements over state-of-the-art graph kernels in terms of predictive performance on various real-world benchmark datasets.

Abstract After more than one decade, Weisfeiler-Lehman graph kernels are still among the most prevalent graph kernels due to their remarkable predictive performance and time complexity. They are based on a fast iterative partitioning of vertices, originally designed for deciding graph isomorphism with one-sided error. The Weisfeiler-Lehman graph kernels retain this idea and compare such labels with respect to equality. This binary valued comparison is, however, arguably too rigid for defining suitable graph kernels for certain graph classes. To overcome this limitation, we propose a generalization of Weisfeiler-Lehman graph kernels which takes into account a more natural and finer grade of similarity between Weisfeiler-Lehman labels than equality. We show that the proposed similarity can be calculated efficiently by means of the Wasserstein distance between certain vectors representing Weisfeiler-Lehman labels. This and other facts give rise to the natural choice of partitioning the vertices with the Wasserstein k-means algorithm. We empirically demonstrate on the Weisfeiler-Lehman subtree kernel, which is one of the most prominent Weisfeiler-Lehman graph kernels, that our generalization significantly outperforms this and other state-of-the-art graph kernels in terms of predictive performance on datasets which contain structurally more complex graphs beyond the typically considered molecular graphs.

Abstract One of the central problems studied in the theory of machine learning is the question of whether, for a given class of hypotheses, it is possible to efficiently find a consistent hypothesis, i.e., one with zero training error. While problems involving convex hypotheses have been extensively studied, the question of whether efficient learning is possible for non-convex hypotheses composed of possibly several disconnected regions is still not well understood. Although it has been shown quite a while ago that efficient learning of weakly convex hypotheses, a parameterized relaxation of convex hypotheses, is possible for the special case of Boolean functions, the question of whether this idea can be developed into a generic paradigm has not yet been studied. In this paper, we provide a positive answer and show that the consistent hypothesis finding problem can indeed be solved in polynomial time for a broad class of weakly convex hypotheses over metric spaces. To this end, we propose a general domain-independent algorithm for finding consistent weakly convex hypotheses and prove sufficient conditions for its efficiency that characterize the corresponding hypothesis classes. To illustrate our general algorithm and its properties, we discuss several non-trivial learning examples to demonstrate how it can be used to efficiently solve the corresponding consistent hypothesis finding problem. Without the weak convexity constraint, these problems are known to be computationally intractable. We then show that the general idea of our algorithm even extends to the extensional case, enabling applications, such as vertex classification in graphs. We prove that using our extended algorithm, the problem can be solved in polynomial time provided the distances in the domain can be computed efficiently.

A possible objective in analyzing trajectories of multiple simultaneously moving objects, such as football players during a game, is to extract and understand the general patterns of coordinated movement in different classes of situations as they develop. For achieving this objective, we propose an approach that includes a combination of query techniques for flexible selection of episodes of situation development, a method for dynamic aggregation of data from selected groups of episodes, and a data structure for representing the aggregates that enables their exploration and use in further analysis. The aggregation, which is meant to abstract general movement patterns, involves construction of new time-homomorphic reference systems owing to iterative application of aggregation operators to a sequence of data selections. As similar patterns may occur at different spatial locations, we also propose constructing new spatial reference systems for aligning and matching movements irrespective of their absolute locations. The approach was tested in application to tracking data from two Bundesliga games of the 2018/2019 season. It enabled detection of interesting and meaningful general patterns of team behaviors in three classes of situations defined by football experts. The experts found the approach and the underlying concepts worth implementing in tools for football analysts.

The word ‘pattern’ frequently appears in the visualisation and visual analytics literature, but what do we mean when we talk about patterns? We propose a practicable definition of the concept of a pattern in a data distribution as a combination of multiple interrelated elements of two or more data components that can be represented and treated as a unified whole. Our theoretical model describes how patterns are made by relationships existing between data elements. Knowing the types of these relationships, it is possible to predict what kinds of patterns may exist. We demonstrate how our model underpins and refines the established fundamental principles of visualisation. The model also suggests a range of interactive analytical operations that can support visual analytics workflows where patterns, once discovered, are explicitly involved in further data analysis.

Decision trees excel at interpretability of their prediction results. To achieve required prediction accuracies, however, often large ensembles of decision trees - random forests - are considered, reducing interpretability due to large size. Additionally, their size slows down inference on modern hardware and restricts their applicability in low-memory embedded devices. We introduce Decision Snippet Features, which are obtained from small subtrees that appear frequently in trained random forests. We subsequently show that linear models on top of these features achieve comparable and sometimes even better predictive performance than the original random forest, while reducing the model size by up to two orders of magnitude.

Frequent subgraph mining, i.e., the identification of relevant patterns in graph databases, is a well-known data mining problem with high practical relevance, since next to summarizing the data, the resulting patterns can also be used to define powerful domain-specific similarity functions for prediction. In recent years, significant progress has been made towards subgraph mining algorithms that scale to complex graphs by focusing on tree patterns and probabilistically allowing a small amount of incompleteness in the result. Nonetheless, the complexity of the pattern matching component used for deciding subtree isomorphism on arbitrary graphs has significantly limited the scalability of existing approaches. In this paper, we adapt sampling techniques from mathematical combinatorics to the problem of probabilistic subtree mining in arbitrary databases of many small to medium-size graphs or a single large graph. By restricting on tree patterns, we provide an algorithm that approximately counts or decides subtree isomorphism for arbitrary transaction graphs in sub-linear time with one-sided error. Our empirical evaluation on a range of benchmark graph datasets shows that the novel algorithm substantially outperforms state-of-the-art approaches both in the task of approximate counting of embeddings in single large graphs and in probabilistic frequent subtree mining in large databases of small to medium sized graphs.

The combinatorial problem of max-sum diversification asks for a maximally diverse subset of a given set of data. Here, we show that it can be expressed as an Ising energy minimization problem. Given this result, max-sum diversification can be solved on adiabatic quantum computers and we present proof of concept simulations which support this claim. This, in turn, suggests that quantum computing might play a role in data mining. We therefore discuss quantum computing in a tutorial like manner and elaborate on its current strengths and weaknesses for data analysis.

In this paper, we present a novel approach to accurately calibrate the kinematic model of a humanoid based on observations of its monocular camera. Our technique estimates the parameters of the complete model, consisting of the joint angle offsets of the whole body including the legs, as well as the camera extrinsic and intrinsic parameters. We cast the parameter estimation as a least-squares optimization problem. In the error function, we consider the residuals between camera observations of end-effector markers and their projections into the image based on the estimate of the calibration parameters. Furthermore, we developed an approach to automatically select a subset of configurations for the calibration process that yields a good trade-off between the number of observations and accuracy. As the experiments with a Nao humanoid show, we achieve an accurate calibration for this low-cost platform. Further, our approach to configuration selection yields substantially better optimization results compared to randomly chosen viable configurations. Hence, our system only requires a reduced number of configurations to achieve accurate results. Our optimization is general and the implementation, which is available online, can easily be applied to different humanoids.

Place-oriented analysis of movement data, i.e., recorded tracks of moving objects, includes finding places of interest in which certain types of movement events occur repeatedly and investigating the temporal distribution of event occurrences in these places and, possibly, other characteristics of the places and links between them. For this class of problems, we propose a visual analytics procedure consisting of four major steps: 1) event extraction from trajectories; 2) extraction of relevant places based on event clustering; 3) spatiotemporal aggregation of events or trajectories; 4) analysis of the aggregated data. All steps can be fulfilled in a scalable way with respect to the amount of the data under analysis; therefore, the procedure is not limited by the size of the computer's RAM and can be applied to very large data sets. We demonstrate the use of the procedure by example of two real-world problems requiring analysis at different spatial scales.

Finding patterns that are interesting to a user in a certain application context is one of the central goals of Data Mining research. Regarding all patterns above a certain frequency threshold as interesting is one way of defining interestingness. In this paper, however, we argue that in many applications, a different notion of interestingness is required in order to be able to capture “long”, and thus particularly informative, patterns that are correspondingly of low frequency. To identify such patterns, our proposed measure of interestingness is based on the degree or strength of closedness of the patterns. We show that (a) indeed this definition selects long interesting patterns that are difficult to identify with frequency-based approaches, and (b) that it selects patterns that are robust against noise and/or dynamic changes. We prove that the family of interesting patterns proposed here forms a closure system and use the corresponding closure operator to design a mining algorithm listing these patterns in amortized quadratic time. In particular, for non-sparse datasets its time complexity is O(nm) per pattern, where n denotes the number of items and m the size of the database. This is equal to the best known time bound for listing ordinary closed frequent sets, which is a special case of our problem. We also report empirical results with real-world datasets.

This article summarizes the results of the workshop on Visualization, Analytics & Spatial Decision Support, which took place at the GIScience conference in September 2006. The discussions at the workshop and analysis of the state of the art have revealed a need in concerted cross‐disciplinary efforts to achieve substantial progress in supporting space‐related decision making. The size and complexity of real‐life problems together with their ill‐defined nature call for a true synergy between the power of computational techniques and the human capabilities to analyze, envision, reason, and deliberate. Existing methods and tools are yet far from enabling this synergy. Appropriate methods can only appear as a result of a focused research based on the achievements in the fields of geovisualization and information visualization, human‐computer interaction, geographic information science, operations research, data mining and machine learning, decision science, cognitive science, and other disciplines. The name ‘Geovisual Analytics for Spatial Decision Support’ suggested for this new research direction emphasizes the importance of visualization and interactive visual interfaces and the link with the emerging research discipline of Visual Analytics. This article, as well as the whole special issue, is meant to attract the attention of scientists with relevant expertise and interests to the major challenges requiring multidisciplinary efforts and to promote the establishment of a dedicated research community where an appropriate range of competences is combined with an appropriate breadth of thinking.

In many applications, unlabelled examples are inexpensive and easy to obtain. Semi-supervised approaches try to utilise such examples to reduce the predictive error. In this paper, we investigate a semi-supervised least squares regression algorithm based on the co-learning approach. Similar to other semi-supervised algorithms, our base algorithm has cubic runtime complexity in the number of unlabelled examples. To be able to handle larger sets of unlabelled examples, we devise a semi-parametric variant that scales linearly in the number of unlabelled examples. Experiments show a significant error reduction by co-regularisation and a large runtime improvement for the semi-parametric approximation. Last but not least, we propose a distributed procedure that can be applied without collecting all data at a single site.

In recent years there has been an increased interest in algorithms that can perform frequent pattern discovery in large databases of graph structured objects. While the frequent connected subgraph mining problem for tree datasets can be solved in incremental polynomial time, it becomes intractable for arbitrary graph databases. Existing approaches have therefore resorted to various heuristic strategies and restrictions of the search space, but have not identified a practically relevant tractable graph class beyond trees. In this paper, we define the class of so called tenuous outerplanar graphs, a strict generalization of trees, develop a frequent subgraph mining algorithm for tenuous outerplanar graphs that works in incremental polynomial time, and evaluate the algorithm empirically on the NCI molecular graph dataset.

Hill-climbing search is the most commonly used search algorithm in ILP systems because it permits the generation of theories in short running times. However, a well known drawback of this greedy search strategy is its myopia. Macro-operators (or macros for short), a recently proposed technique to reduce the search space explored by exhaustive search, can also be argued to reduce the myopia of hill-climbing search by automatically performing a variable-depth look-ahead in the search space. Surprisingly, macros have not been employed in a greedy learner. In this paper, we integrate macros into a hill-climbing learner. In a detailed comparative study in several domains, we show that indeed a hill-climbing learner using macros performs significantly better than current state-of-the-art systems involving other techniques for reducing myopia, such as fixed-depth look-ahead, template-based look-ahead, beam-search, or determinate literals. In addition, macros, in contrast to some of the other approaches, can be computed fully automatically and do not require user involvement nor special domain properties such as determinacy.

S.158-167

Given the enormous amounts of information available only in unstructured or semi-structured textual documents, tools for information extraction (IE) have become enormously important. IE tools identify the relevant information in such documents and convert it into a structured format such as a database or an XML document. While first IE algorithms were hand-crafted sets of rules, researchers soon turned to learning extraction rules from hand-labeled documents. Unfortunately, rule-based approaches sometimes fail to provide the necessary robustness against the inherent variability of document, structure, which has led to the recent interest in using hidden Markov models (HMMs). By using additional unlabeled documents as they are usually readily available in most applications, we can perform active learning of HMMs. The idea of active learning algorithms is to identify unlabeled observations that would be most useful when labeled by the user. Such algorithms are known for classification, clustering, and regression; we present the first algorithm for active learning of hidden Markov models.

Many discovery problems, e.g., subgroup or association rule discovery, can naturally be cast as n-best hypothesis problems where the goal is to nd the n hypotheses from a given hypothesis space that score best according to a given utility function. We present a sampling algorithm that solves this problem by issuing a small number of database queries while guaranteeing precise bounds on condence and quality of solutions. Known sampling algorithms assume that the utility be the average (over the examples) of some function, which is not the case for many frequently used utility functions. We show that our algorithm works for all utilities that can be estimated with bounded error. We provide such error bounds and resulting worst-case sample bounds for some of the most frequently used utilities, and prove that there is no sampling algorithm for another popular class of utility functions. The algorithm is sequential in the sense that it starts to return (or discard) hypotheses that already...