The goal of inductive program synthesis is for a machine to automatically generate a program from user-supplied examples. A key underlying assumption is that humans can provide sufficient examples to teach a concept to a machine. To evaluate the validity of this assumption, we conduct a study where human participants provide examples for six programming concepts, such as finding the maximum element of a list. We evaluate the generalisation performance of five program synthesis systems trained on input-output examples (i) from a human group, (ii) from a gold standard set, and (iii) randomly sampled. Our results suggest that human-provided examples are typically insufficient for a program synthesis system to learn an accurate program.

Context: Computer science, even now, attracts a small number of women, and the proportion of women in the field decreases through advancing career stages. Consequently, few women progress to Ph.D. studies in computer science after completing master’s studies. Empowering women at this stage in their careers is essential, not just for equality reasons, but to unlock untapped potential for society, industry and academia. Objective: This paper aims to identify students’ career assumptions and information related to Ph.D. studies focused on gender-based differences. We propose a program to inform female master students about Ph.D. studies that explains the process, clarifies misconceptions, and alleviates concerns. Method: An extensive survey was conducted to identify factors that encourage and discourage students from undertaking Ph.D. studies. The analysis identified statistically significant differences between those who undertook Ph.D. studies and those who did not, as well as statistically significant gender differences. A catalogue of questions to initiate discussions with potential Ph.D. students which allowed them to explore these factors was developed. These were structured into a Women’s Career Lunch program where students can explore and discuss the benefits of Ph.D. study. Results: Encouraging factors towards Ph.D. study include interest and confidence in research arising from a research involvement during earlier studies; enthusiasm for and self-confidence in computer science in addition to an interest in an academic career; encouragement from external sources; and a positive perception towards Ph.D. studies which can involve achieving personal goals. Discouraging factors include uncertainty and lack of knowledge of the Ph.D. process, a perception of lower job flexibility, and the requirement for long-term commitment. Gender differences highlighted that female students who pursue a Ph.D. have less confidence in their technical skills than males but a higher preference for interdisciplinary areas. Female students are less inclined than males to perceive the industry as offering better job opportunities and more flexible career paths than academia. Conclusions: The insights collected from the survey facilitated the development of a questions catalogue structured into the Women Career Lunch program to help students make a more informed decision concerning whether they should pursue a Ph.D. in computer science. Localised versions of this program, in 8 languages, were created to support its adoption in different countries and assist in mitigating the female under-representation challenge.

Abstract In this paper, we introduce a post-hoc and local explainable AI method tailored for Knowledge Graph Embedding (KGE) models. These models are essential to Knowledge Graph Completion yet criticized for their black-box nature. Despite their success in capturing the semantics of knowledge graphs through high-dimensional latent representations, their inherent complexity poses substantial challenges to explainability. While existing methods like Kelpie use resource-intensive perturbation to explain KGE models, our approach directly decodes the latent representations encoded by KGE models, leveraging the smoothness of the embeddings, which follows the principle that similar embeddings reflect similar behaviours within the Knowledge Graph, meaning that nodes are similarly embedded because their graph neighbourhood looks similar. This principle is commonly referred to as smoothness. By identifying symbolic structures, in the form of triples, within the subgraph neighborhoods of similarly embedded entities, our method identifies the statistical regularities on which the models rely and translates these insights into human-understandable symbolic rules and facts. This bridges the gap between the abstract representations of KGE models and their predictive outputs, providing clear and interpretable insights. The contributions include a novel post-hoc and local explainable AI method for KGE models, which provides immediate and faithful explanations without retraining, thereby facilitating real-time application on large-scale knowledge graphs. The method’s flexibility enables the generation of rule-based, instance-based, and analogy-based explanations, meeting diverse user needs. Extensive evaluations show the effectiveness of our approach in delivering faithful and well-localized explanations, enhancing the transparency and trustworthiness of KGE models.

Abstract We introduce FMC-Net, a facial expression recognition (FER) framework that leverages the hierarchical relationship between discrete facial muscle movements, known as Action Units (AUs), and Facial Expressions (FEs) by integrating two complementary constraint layers. This framework couples data-driven learning with psychology-grounded structure. First, a training-time correlation constraint aligns the two tasks within a multi-task network by softly regularizing a target statistical relationship. This can improve sample efficiency and generalization, particularly under limited or biased data. Second, an inference-time fuzzy rule layer maps the networks probabilistic AU predictions to FEs using compact, human-editable from psychological research, yielding transparent, per-decision attributions. An ensemble then combines the model and rule-based pathways and exposes a disagreement-based risk score for human-in-the-loop triage. This two-layer constraint integration addresses the limitations of single-mechanism approaches: training-time constraints shape the learned representations but lack case-wise transparency, while inference-time rules explain decisions but cannot improve the underlying features. Experiments across diverse datasets, including in-the-wild video and cross-dataset evaluation, validate our approach. Constraint-guided training consistently produces models that outperform competitive baselines, while the rule-based pathway can provide transparency and actionable risk signals towards reliable deployment. The proposed methodology is also generalizable to other machine learning tasks with interdependent outputs.

We introduce FruitNeRF, a unified novel fruit counting framework that leverages state-of-the-art view synthesis methods to count any fruit type directly in 3D. Our framework takes an unordered set of posed images captured by a monocular camera and segments fruit in each image. To make our system independent of the fruit type, we employ a foundation model that generates binary segmentation masks for any fruit. Utilizing both modalities, RGB and semantic, we train a semantic neural radiance field. Through uniform volume sampling of the implicit Fruit Field, we obtain fruit-only point clouds. By applying cascaded clustering on the extracted point cloud, our approach achieves precise fruit count. The use of neural radiance fields provides significant advantages over conventional methods such as object tracking or optical flow, as the counting itself is lifted into 3D. Our method prevents double counting fruit and avoids counting irrelevant fruit. We evaluate our methodology using both real-world and synthetic datasets. The real-world dataset consists of three apple trees with manually counted ground truths, a benchmark apple dataset with one row and ground truth fruit location, while the synthetic dataset comprises various fruit types including apple, plum, lemon, pear, peach, and mango. Additionally, we assess the performance of fruit counting using the foundation model compared to a U-Net.

Abstract The growing number of applications of machine learning and data mining in many domains—from agriculture to business, education, industrial manufacturing, and medicine—gave rise to new requirements for how to inspect and control the learned models. The research domain of explainable artificial intelligence (XAI) has been newly established with a strong focus on methods being applied post-hoc on black-box models. As an alternative, the use of interpretable machine learning methods has been considered—where the learned models are white-box ones. Black-box models can be characterized as representing implicit knowledge—typically resulting from statistical and neural approaches of machine learning, while white-box models are explicit representations of knowledge—typically resulting from rule-learning approaches. In this introduction to the special issue on ‘Explainable and Interpretable Machine Learning and Data Mining’ we propose to bring together both perspectives, pointing out commonalities and discussing possibilities to integrate them.

Abstract Research in the field of explainable artificial intelligence has produced a vast amount of visual explanation methods for deep learning-based image classification in various domains of application. However, there is still a lack of domain-specific evaluation methods to assess an explanation’s quality and a classifier’s performance with respect to domain-specific requirements. In particular, evaluation methods could benefit from integrating human expertise into quality criteria and metrics. Such domain-specific evaluation methods can help to assess the robustness of deep learning models more precisely. In this paper, we present an approach for domain-specific evaluation of visual explanation methods in order to enhance the transparency of deep learning models and estimate their robustness accordingly. As an example use case, we apply our framework to facial expression recognition. We can show that the domain-specific evaluation is especially beneficial for challenging use cases such as facial expression recognition and provides application-grounded quality criteria that are not covered by standard evaluation methods. Our comparison of the domain-specific evaluation method with standard approaches thus shows that the quality of the expert knowledge is of great importance for assessing a model’s performance precisely.

Abstract The topic of comprehensibility of machine-learned theories has recently drawn increasing attention. Inductive logic programming uses logic programming to derive logic theories from small data based on abduction and induction techniques. Learned theories are represented in the form of rules as declarative descriptions of obtained knowledge. In earlier work, the authors provided the first evidence of a measurable increase in human comprehension based on machine-learned logic rules for simple classification tasks. In a later study, it was found that the presentation of machine-learned explanations to humans can produce both beneficial and harmful effects in the context of game learning. We continue our investigation of comprehensibility by examining the effects of the ordering of concept presentations on human comprehension. In this work, we examine the explanatory effects of curriculum order and the presence of machine-learned explanations for sequential problem-solving. We show that (1) there exist tasks A and B such that learning A before learning B results in better comprehension for humans in comparison to learning B before learning A and (2) there exist tasks A and B such that the presence of explanations when learning A contributes to improved human comprehension when subsequently learning B. We propose a framework for the effects of sequential teaching on comprehension based on an existing definition of comprehensibility and provide evidence for support from data collected in human trials. Our empirical study involves curricula that teach novices the merge sort algorithm. Our results show that sequential teaching of concepts with increasing complexity (a) has a beneficial effect on human comprehension and (b) leads to human re-discovery of divide-and-conquer problem-solving strategies, and (c) allows adaptations of human problem-solving strategy with better performance when machine-learned explanations are also presented.

We present ManuKnowVis, the result of a design study, in which we contextualize data from multiple knowledge repositories of a manufacturing process for battery modules used in electric vehicles. In data-driven analyses of manufacturing data, we observed a discrepancy between two stakeholder groups involved in serial manufacturing processes: Knowledge providers (e.g., engineers) have domain knowledge about the manufacturing process but have difficulties in implementing data-driven analyses. Knowledge consumers (e.g., data scientists) have no first-hand domain knowledge but are highly skilled in performing data-driven analyses. ManuKnowVis bridges the gap between providers and consumers and enables the creation and completion of manufacturing knowledge. We contribute a multi-stakeholder design study, where we developed ManuKnowVis in three main iterations with consumers and providers from an automotive company. The iterative development led us to a multiple linked view tool, in which, on the one hand, providers can describe and connect individual entities (e.g., stations or produced parts) of the manufacturing process based on their domain knowledge. On the other hand, consumers can leverage this enhanced data to better understand complex domain problems, thus, performing data analyses more efficiently. As such, our approach directly impacts the success of data-driven analyses from manufacturing data. To demonstrate the usefulness of our approach, we carried out a case study with seven domain experts, which demonstrates how providers can externalize their knowledge and consumers can implement data-driven analyses more efficiently.

Machine learning based image classification algorithms, such as deep neural network approaches, will be increasingly employed in critical settings such as quality control in industry, where transparency and comprehensibility of decisions are crucial. Therefore, we aim to extend the defect detection task towards an interactive human-in-the-loop approach that allows us to integrate rich background knowledge and the inference of complex relationships going beyond traditional purely data-driven approaches. We propose an approach for an interactive support system for classifications in an industrial quality control setting that combines the advantages of both (explainable) knowledge-driven and data-driven machine learning methods, in particular inductive logic programming and convolutional neural networks, with human expertise and control. The resulting system can assist domain experts with decisions, provide transparent explanations for results, and integrate feedback from users; thus reducing workload for humans while both respecting their expertise and without removing their agency or accountability.

Neural networks are widely adopted, yet the integration of domain knowledge is still underutilized. We propose to integrate domain knowledge about co-occurring facial movements as a constraint in the loss function to enhance the training of neural networks for affect recognition. As the co-ccurrence patterns tend to be similar across datasets, applying our method can lead to a higher generalizability of models and a lower risk of overfitting. We demonstrate this by showing performance increases in cross-dataset testing for various datasets. We also show the applicability of our method for calibrating neural networks to different facial expressions.

Abstract In recent research, human-understandable explanations of machine learning models have received a lot of attention. Often explanations are given in form of model simplifications or visualizations. However, as shown in cognitive science as well as in early AI research, concept understanding can also be improved by the alignment of a given instance for a concept with a similar counterexample. Contrasting a given instance with a structurally similar example which does not belong to the concept highlights what characteristics are necessary for concept membership. Such near misses have been proposed by Winston (Learning structural descriptions from examples, 1970) as efficient guidance for learning in relational domains. We introduce an explanation generation algorithm for relational concepts learned with Inductive Logic Programming (GeNME). The algorithm identifies near miss examples from a given set of instances and ranks these examples by their degree of closeness to a specific positive instance. A modified rule which covers the near miss but not the original instance is given as an explanation. We illustrateGeNMEwith the well-known family domain consisting of kinship relations, the visual relational Winston arches domain, and a real-world domain dealing with file management. We also present a psychological experiment comparing human preferences of rule-based, example-based, and near miss explanations in the family and the arches domains.

In this design study, we present IRVINE, a Visual Analytics (VA) system, which facilitates the analysis of acoustic data to detect and understand previously unknown errors in the manufacturing of electrical engines. In serial manufacturing processes, signatures from acoustic data provide valuable information on how the relationship between multiple produced engines serves to detect and understand previously unknown errors. To analyze such signatures, IRVINE leverages interactive clustering and data labeling techniques, allowing users to analyze clusters of engines with similar signatures, drill down to groups of engines, and select an engine of interest. Furthermore, IRVINE allows to assign labels to engines and clusters and annotate the cause of an error in the acoustic raw measurement of an engine. Since labels and annotations represent valuable knowledge, they are conserved in a knowledge database to be available for other stakeholders. We contribute a design study, where we developed IRVINE in four main iterations with engineers from a company in the automotive sector. To validate IRVINE, we conducted a field study with six domain experts. Our results suggest a high usability and usefulness of IRVINE as part of the improvement of a real-world manufacturing process. Specifically, with IRVINE domain experts were able to label and annotate produced electrical engines more than 30% faster.

This work focuses on computationally efficient difference metrics of time series and compares two different unsupervised methods for anomaly classification. It takes place in the domain of hardware systems testing for reliability, where several structurally identical devices are tested at the same time with a load expected in their lifetime use. The devices perform different maneuvers in predefined testing cycles. It is possible that rare, unexpected system defects appear. They often show up in the measured data signals of the system, for example as a decrease in the output pressure of a pump. Due to the intended aging of the parts under load, the measured data also exhibits a concept drift, i.e. a shift in the data distribution. It is of interest to detect anomalous behavior as early as possible to reduce cost, save time and enable accurate root-cause-analysis. We formulate this problem as an anomaly detection task on periodic multivariate time series data. Experiments are evaluated using an open access hydraulic test bench data set by Helwig et al. [1]. The method's performance under concept drift is tested by simulating an aging system using the same data set. We find that Mean Squared Error towards the median in combination with the Modified z-Score is the most robust method for this use case. The solution can be applied from the beginning of a hardware testing cycle. The computations are intuitive to understand, and the classification results can be visualized for better interpretability and plausibility analysis.

Abstract Given the recent successes of Deep Learning in AI there has been increased interest in the role and need for explanations in machine learned theories. A distinct notion in this context is that of Michie’s definition of ultra-strong machine learning (USML). USML is demonstrated by a measurable increase in human performance of a task following provision to the human of a symbolic machine learned theory for task performance. A recent paper demonstrates the beneficial effect of a machine learned logic theory for a classification task, yet no existing work to our knowledge has examined the potential harmfulness of machine’s involvement for human comprehension during learning. This paper investigates the explanatory effects of a machine learned theory in the context of simple two person games and proposes a framework for identifying the harmfulness of machine explanations based on the Cognitive Science literature. The approach involves a cognitive window consisting of two quantifiable bounds and it is supported by empirical evidence collected from human trials. Our quantitative and qualitative results indicate that human learning aided by a symbolic machine learned theory which satisfies a cognitive window has achieved significantly higher performance than human self learning. Results also demonstrate that human learning aided by a symbolic machine learned theory that fails to satisfy this window leads to significantly worse performance than unaided human learning.

Pain sensation is essential for survival, since it draws attention to physical threat to the body. Pain assessment is usually done through self-reports. However, self-assessment of pain is not available in the case of noncommunicative patients, and therefore, observer reports should be relied upon. Observer reports of pain could be prone to errors due to subjective biases of observers. Moreover, continuous monitoring by humans is impractical. Therefore, automatic pain detection technology could be deployed to assist human caregivers and complement their service, thereby improving the quality of pain management, especially for noncommunicative patients. Facial expressions are a reliable indicator of pain, and are used in all observer-based pain assessment tools. Following the advancements in automatic facial expression analysis, computer vision researchers have tried to use this technology for developing approaches for automatically detecting pain from facial expressions. This paper surveys the literature published in this field over the past decade, categorizes it, and identifies future research directions. The survey covers the pain datasets used in the reviewed literature, the learning tasks targeted by the approaches, the features extracted from images and image sequences to represent pain-related information, and finally, the machine learning methods used.

During the 1980s Michie defined Machine Learning in terms of two orthogonal axes of performance: predictive accuracy and comprehensibility of generated hypotheses. Since predictive accuracy was readily measurable and comprehensibility not so, later definitions in the 1990s, such as Mitchell’s, tended to use a one-dimensional approach to Machine Learning based solely on predictive accuracy, ultimately favouring statistical over symbolic Machine Learning approaches. In this paper we provide a definition of comprehensibility of hypotheses which can be estimated using human participant trials. We present two sets of experiments testing human comprehensibility of logic programs. In the first experiment we test human comprehensibility with and without predicate invention. Results indicate comprehensibility is affected not only by the complexity of the presented program but also by the existence of anonymous predicate symbols. In the second experiment we directly test whether any state-of-the-art ILP systems are ultra-strong learners in Michie’s sense, and select the Metagol system for use in humans trials. Results show participants were not able to learn the relational concept on their own from a set of examples but they were able to apply the relational definition provided by the ILP system correctly. This implies the existence of a class of relational concepts which are hard to acquire for humans, though easy to understand given an abstract explanation. We believe improved understanding of this class could have potential relevance to contexts involving human learning, teaching and verbal interaction.

While some computational models of intelligence test problems were proposed throughout the second half of the XXth century, in the first years of the XXIst century we have seen an increasing number of computer systems being able to score well on particular intelligence test tasks. However, despitethis increasing trend there has been no general account of all these works in terms of how theyrelate to each other and what their real achievements are. In this paper, we provide some insighton these issues by giving a comprehensive account of about thirty computer models, from the 1960sto nowadays, and their relationships, focussing on the range of intelligence test tasks they address, thepurpose of the models, how general or specialised these models are, the AI techniques they use in eachcase, their comparison with human performance, and their evaluation of item difficulty.

In many domains of computer vision, such as medical imaging and facial image analysis, it is necessary to combine shape (geometric) and appearance (texture) information. In this paper, we describe a method for combining geometric and texture-based evidence for facial actions within a Kalman filter framework. The geometric evidence is provided by a face alignment method. The texture-based evidence is provided by a set of Support Vector Machines (SVM) for various Action Units (AU). The proposed method is a practical solution to the problem of fusing categorical probabilities within a Kalman filter based state estimation framework. A first performance evaluation on upper face AUs demonstrates the practical applicability of the proposed fusion method. The method is applicable to arbitrary imaging domains, apart from facial action estimation.

While some computational models of intelligence test problems were proposed throughout the second half of the XXth century, in the first years of the XXIst century we have seen an increasing number of computer systems being able to score well on particular intelligence test tasks. However, despite this increasing trend there has been no general account of all these works in terms of how they relate to each other and what their real achievements are. Also, there is poor understanding about what intelligence tests measure in machines, whether they are useful to evaluate AI systems, whether they are really challenging problems, and whether they are useful to understand (human) intelligence. In this paper, we provide some insight on these issues, in the form of nine specific questions, by giving a comprehensive account of about thirty computer models, from the 1960s to nowadays, and their relationships, focussing on the range of intelligence test tasks they address, the purpose of the models, how general or specialised these models are, the AI techniques they use in each case, their comparison with human performance, and their evaluation of item difficulty. As a conclusion, these tests and the computer models attempting them show that AI is still lacking general techniques to deal with a variety of problems at the same time. Nonetheless, a renewed attention on these problems and a more careful understanding of what intelligence tests offer for AI may help build new bridges between psychometrics, cognitive science, and AI; and may motivate new kinds of problem repositories.

We present an extension to a current approach to inductive programming (IGOR2), that is, learning (recursive) programs from incomplete specifications such as input/outout examples. IGOR2 uses an analytical, example-driven strategy for generalization. We extend the set of IGOR2's refinement operators by a further operator – identification of higher-order schemes – and can show that this extension does improve speed as well as scope.

We describe an approach to the inductive synthesis of recursive equations from input/output-examples which is based on the classical two-step approach to induction of functional Lisp programs of Summers (1977). In a first step, I/O-examples are rewritten to traces which explain the outputs given the respective inputs based on a datatype theory. These traces can be integrated into one conditional expression which represents a non-recursive program. In a second step, this initial program term is generalized into recursive equations by searching for syntactical regularities in the term. Our approach extends the classical work in several aspects. The most important extensions are that we are able to induce a set of recursive equations in one synthesizing step, the equations may contain more than one recursive call, and additionally needed parameters are automatically introduced.