Research Papers

Explanation methods and their evaluation have become a significant issue in explainable artificial intelligence (XAI) due to the recent surge of opaque AI models in decision support systems (DSS). Since the most accurate AI models are opaque with low transparency and comprehensibility, explanations are essential for bias detection and control of uncertainty. There are a plethora of criteria to choose from when evaluating explanation method quality. However, since existing criteria focus on evaluating single explanation methods, it is not obvious how to compare the quality of different methods. This lack of consensus creates a critical shortage of rigour in the field, although little is written about comparative evaluations of explanation methods. In this paper, we have conducted a semi-systematic meta-survey over fifteen literature surveys covering the evaluation of explainability to identify existing criteria usable for comparative evaluations of explanation methods. The main contribution in the paper is the suggestion to use appropriate trust as a criterion to measure the outcome of the subjective evaluation criteria and consequently make comparative evaluations possible. We also present a model of explanation quality aspects. In the model, criteria with similar definitions are grouped and related to three identified aspects of quality; model, explanation, and user. We also notice four commonly accepted criteria (groups) in the literature, covering all aspects of explanation quality: Performance, appropriate trust, explanation satisfaction, and fidelity. We suggest the model be used as a chart for comparative evaluations to create more generalisable research in explanation quality.

Trust is often cited as an essential criterion for the effective use and real-world deployment of AI. Researchers argue that AI should be more transparent to increase trust, making transparency one of the main goals of XAI. Nevertheless, empirical research on this topic is inconclusive regarding the effect of transparency on trust. An explanation for this ambiguity could be that trust is operationalized differently within XAI. In this position paper, we advocate for a clear distinction between behavioral (objective) measures of reliance and attitudinal (subjective) measures of trust. However, researchers sometimes appear to use behavioral measures when intending to capture trust, although attitudinal measures would be more appropriate. Based on past research, we emphasize that there are sound theoretical reasons to keep trust and reliance separate. Properly distinguishing these two concepts provides a more comprehensive understanding of how transparency affects trust and reliance, benefiting future XAI research.

Feature selection of high-dimensional labeled data with limited observations is critical for making powerful predictive modeling accessible, scalable, and interpretable for domain experts. Spectroscopy data, which records the interaction between matter and electromagnetic radiation, particularly holds a lot of information in a single sample. Since acquiring such high-dimensional data is a complex task, it is crucial to exploit the best analytical tools to extract necessary information. In this paper, we investigate the most commonly used feature selection techniques and introduce applying recent explainable AI techniques to interpret the prediction outcomes of high-dimensional and limited spectral data. Interpretation of the prediction outcome is beneficial for the domain experts as it ensures the transparency and faithfulness of the ML models to the domain knowledge. Due to the instrument resolution limitations, pinpointing important regions of the spectroscopy data creates a pathway to optimize the data collection process through the miniaturization of the spectrometer device. Reducing the device size and power and therefore cost is a requirement for the real-world deployment of such a sensor-to-prediction system as a whole. We specifically design three different scenarios to ensure that the evaluation of ML models is robust for the real-time practice of the developed methodologies and to uncover the hidden effect of noise sources on the final outcome.

Modern AI models contain much of human knowledge, yet understanding of their internal representation of this knowledge remains elusive. Characterizing the structure and properties of this representation will lead to improvements in model capabilities and development of effective safeguards. Building on recent advances in feature learning, we develop an effective, scalable approach for extracting linear representations of general concepts in large-scale AI models (language models, vision-language models, and reasoning models). We show how these representations enable model steering, through which we expose vulnerabilities, mitigate misaligned behaviors, and improve model capabilities. Additionally, we demonstrate that concept representations are remarkably transferable across human languages and combinable to enable multi-concept steering. Through quantitative analysis across hundreds of concepts, we find that newer, larger models are more steerable and steering can improve model capabilities beyond standard prompting. We show how concept representations are effective for monitoring misaligned content (hallucinations, toxic content). We demonstrate that predictive models built using concept representations are more accurate for monitoring misaligned content than using models that judge outputs directly. Together, our results illustrate the power of using internal representations to map the knowledge in AI models, advance AI safety, and improve model capabilities.

This paper investigates the relationship between law and eXplainable Artificial Intelligence (XAI). While there is much discussion about the AI Act, for which the trilogue of the European Parliament, Council and Commission recently concluded, other areas of law seem underexplored. This paper focuses on European (and in part German) law, although with international concepts and regulations such as fiduciary plausibility checks, the General Data Protection Regulation (GDPR), and product safety and liability. Based on XAI-taxonomies, requirements for XAI-methods are derived from each of the legal bases, resulting in the conclusion that each legal basis requires different XAI properties and that the current state of the art does not fulfill these to full satisfaction, especially regarding the correctness (sometimes called fidelity) and confidence estimates of XAI-methods. Published in the Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society https://doi.org/10.1609/aies.v7i1.31648 .

Understanding the (Extra-)Ordinary: Validating Deep Model Decisions with Prototypical Concept-based Explanations

Maximilian Dreyer, Reduan Achtibat, Wojciech Samek +1

Ensuring both transparency and safety is critical when deploying Deep Neural Networks (DNNs) in high-risk applications, such as medicine. The field of explainable AI (XAI) has proposed various methods to comprehend the decision-making processes of opaque DNNs. However, only few XAI methods are suitable of ensuring safety in practice as they heavily rely on repeated labor-intensive and possibly biased human assessment. In this work, we present a novel post-hoc concept-based XAI framework that conveys besides instance-wise (local) also class-wise (global) decision-making strategies via prototypes. What sets our approach apart is the combination of local and global strategies, enabling a clearer understanding of the (dis-)similarities in model decisions compared to the expected (prototypical) concept use, ultimately reducing the dependence on human long-term assessment. Quantifying the deviation from prototypical behavior not only allows to associate predictions with specific model sub-strategies but also to detect outlier behavior. As such, our approach constitutes an intuitive and explainable tool for model validation. We demonstrate the effectiveness of our approach in identifying out-of-distribution samples, spurious model behavior and data quality issues across three datasets (ImageNet, CUB-200, and CIFAR-10) utilizing VGG, ResNet, and EfficientNet architectures. Code is available on https://github.com/maxdreyer/pcx.

Artificial intelligence (AI) models trained using medical images for clinical tasks often exhibit bias in the form of disparities in performance between subgroups. Since not all sources of biases in real-world medical imaging data are easily identifiable, it is challenging to comprehensively assess how those biases are encoded in models, and how capable bias mitigation methods are at ameliorating performance disparities. In this article, we introduce a novel analysis framework for systematically and objectively investigating the impact of biases in medical images on AI models. We developed and tested this framework for conducting controlled in silico trials to assess bias in medical imaging AI using a tool for generating synthetic magnetic resonance images with known disease effects and sources of bias. The feasibility is showcased by using three counterfactual bias scenarios to measure the impact of simulated bias effects on a convolutional neural network (CNN) classifier and the efficacy of three bias mitigation strategies. The analysis revealed that the simulated biases resulted in expected subgroup performance disparities when the CNN was trained on the synthetic datasets. Moreover, reweighing was identified as the most successful bias mitigation strategy for this setup, and we demonstrated how explainable AI methods can aid in investigating the manifestation of bias in the model using this framework. Developing fair AI models is a considerable challenge given that many and often unknown sources of biases can be present in medical imaging datasets. In this work, we present a novel methodology to objectively study the impact of biases and mitigation strategies on deep learning pipelines, which can support the development of clinical AI that is robust and responsible.

Load shapes derived from smart meter data are frequently employed to analyze daily energy consumption patterns, particularly in the context of applications like Demand Response (DR). Nevertheless, one of the most important challenges to this endeavor lies in identifying the most suitable consumer clusters with similar consumption behaviors. In this paper, we present a novel machine learning based framework in order to achieve optimal load profiling through a real case study, utilizing data from almost 5000 households in London. Four widely used clustering algorithms are applied specifically K-means, K-medoids, Hierarchical Agglomerative Clustering and Density-based Spatial Clustering. An empirical analysis as well as multiple evaluation metrics are leveraged to assess those algorithms. Following that, we redefine the problem as a probabilistic classification one, with the classifier emulating the behavior of a clustering algorithm,leveraging Explainable AI (xAI) to enhance the interpretability of our solution. According to the clustering algorithm analysis the optimal number of clusters for this case is seven. Despite that, our methodology shows that two of the clusters, almost 10\% of the dataset, exhibit significant internal dissimilarity and thus it splits them even further to create nine clusters in total. The scalability and versatility of our solution makes it an ideal choice for power utility companies aiming to segment their users for creating more targeted Demand Response programs.

Machine learning offers great potential for automated prediction of post-stroke symptoms and their response to rehabilitation. Major challenges for this endeavour include the very high dimensionality of neuroimaging data, the relatively small size of the datasets available for learning, and how to effectively combine neuroimaging and tabular data (e.g. demographic information and clinical characteristics). This paper evaluates several solutions based on two strategies. The first is to use 2D images that summarise MRI scans. The second is to select key features that improve classification accuracy. Additionally, we introduce the novel approach of training a convolutional neural network (CNN) on images that combine regions-of-interest extracted from MRIs, with symbolic representations of tabular data. We evaluate a series of CNN architectures (both 2D and a 3D) that are trained on different representations of MRI and tabular data, to predict whether a composite measure of post-stroke spoken picture description ability is in the aphasic or non-aphasic range. MRI and tabular data were acquired from 758 English speaking stroke survivors who participated in the PLORAS study. The classification accuracy for a baseline logistic regression was 0.678 for lesion size alone, rising to 0.757 and 0.813 when initial symptom severity and recovery time were successively added. The highest classification accuracy 0.854 was observed when 8 regions-of-interest was extracted from each MRI scan and combined with lesion size, initial severity and recovery time in a 2D Residual Neural Network.Our findings demonstrate how imaging and tabular data can be combined for high post-stroke classification accuracy, even when the dataset is small in machine learning terms. We conclude by proposing how the current models could be improved to achieve even higher levels of accuracy using images from hospital scanners.

Recent advancements have significantly improved the efficiency and effectiveness of deep learning methods for imagebased remote sensing tasks. However, the requirement for large amounts of labeled data can limit the applicability of deep neural networks to existing remote sensing datasets. To overcome this challenge, fewshot learning has emerged as a valuable approach for enabling learning with limited data. While previous research has evaluated the effectiveness of fewshot learning methods on satellite based datasets, little attention has been paid to exploring the applications of these methods to datasets obtained from UAVs, which are increasingly used in remote sensing studies. In this review, we provide an up to date overview of both existing and newly proposed fewshot classification techniques, along with appropriate datasets that are used for both satellite based and UAV based data. Our systematic approach demonstrates that fewshot learning can effectively adapt to the broader and more diverse perspectives that UAVbased platforms can provide. We also evaluate some SOTA fewshot approaches on a UAV disaster scene classification dataset, yielding promising results. We emphasize the importance of integrating XAI techniques like attention maps and prototype analysis to increase the transparency, accountability, and trustworthiness of fewshot models for remote sensing. Key challenges and future research directions are identified, including tailored fewshot methods for UAVs, extending to unseen tasks like segmentation, and developing optimized XAI techniques suited for fewshot remote sensing problems. This review aims to provide researchers and practitioners with an improved understanding of fewshot learnings capabilities and limitations in remote sensing, while highlighting open problems to guide future progress in efficient, reliable, and interpretable fewshot methods.

Interpretability is in the Mind of the Beholder: A Causal Framework for Human-interpretable Representation Learning

Emanuele Marconato, Andrea Passerini, Stefano Teso

Focus in Explainable AI is shifting from explanations defined in terms of low-level elements, such as input features, to explanations encoded in terms of interpretable concepts learned from data. How to reliably acquire such concepts is, however, still fundamentally unclear. An agreed-upon notion of concept interpretability is missing, with the result that concepts used by both post-hoc explainers and concept-based neural networks are acquired through a variety of mutually incompatible strategies. Critically, most of these neglect the human side of the problem: a representation is understandable only insofar as it can be understood by the human at the receiving end. The key challenge in Human-interpretable Representation Learning (HRL) is how to model and operationalize this human element. In this work, we propose a mathematical framework for acquiring interpretable representations suitable for both post-hoc explainers and concept-based neural networks. Our formalization of HRL builds on recent advances in causal representation learning and explicitly models a human stakeholder as an external observer. This allows us to derive a principled notion of alignment between the machine representation and the vocabulary of concepts understood by the human. In doing so, we link alignment and interpretability through a simple and intuitive name transfer game, and clarify the relationship between alignment and a well-known property of representations, namely disentanglment. We also show that alignment is linked to the issue of undesirable correlations among concepts, also known as concept leakage, and to content-style separation, all through a general information-theoretic reformulation of these properties. Our conceptualization aims to bridge the gap between the human and algorithmic sides of interpretability and establish a stepping stone for new research on human-interpretable representations.

Explainable AI applications in the Medical Domain: a systematic review

Nicoletta Prentzas, Antonis Kakas, Constantinos S. Pattichis

Artificial Intelligence in Medicine has made significant progress with emerging applications in medical imaging, patient care, and other areas. While these applications have proven successful in retrospective studies, very few of them were applied in practice.The field of Medical AI faces various challenges, in terms of building user trust, complying with regulations, using data ethically.Explainable AI (XAI) aims to enable humans understand AI and trust its results. This paper presents a literature review on the recent developments of XAI solutions for medical decision support, based on a representative sample of 198 articles published in recent years. The systematic synthesis of the relevant articles resulted in several findings. (1) model-agnostic XAI techniques were mostly employed in these solutions, (2) deep learning models are utilized more than other types of machine learning models, (3) explainability was applied to promote trust, but very few works reported the physicians participation in the loop, (4) visual and interactive user interface is more useful in understanding the explanation and the recommendation of the system. More research is needed in collaboration between medical and AI experts, that could guide the development of suitable frameworks for the design, implementation, and evaluation of XAI solutions in medicine.