Umeå University's logo

Current eXplainable AI (XAI) methods tend to provide only one-way explanations, limiting user interaction and contextual adaptation. Social Explainable AI (sXAI) addresses this by enabling interactive, co-constructed explanations. This paper demonstrates how Contextual Importance and Utility (CIU) and Knowledge Graphs (KGs) can generate structured and context-aware explanations. We present a proof-of-concept implementation that shows how KGs facilitate dynamic dialogues, making sXAI practical. Our findings highlight the potential of CIU and KGs in creating more user-centered, interactive explainability frameworks.

Artificial Intelligence (AI) is transforming industries, offering new opportunities to manage and enhance innovation. However, these advancements bring significant challenges for scientists and businesses, with one of the most critical being the ‘trustworthiness” of AI systems. A key requirement of trustworthiness is transparency, closely linked to explicability. Consequently, the exponential growth of eXplainable AI (XAI) has led to the development of numerous methods and metrics for explainability. Nevertheless, this has resulted in a lack of standardized and formal definitions for fundamental XAI properties (e.g., what do soundness, completeness, and faithfulness of an explanation entail? How is the stability of an XAI method defined?). This lack of consensus makes it difficult for XAI practitioners to establish a shared foundation, thereby impeding the effective benchmarking of XAI methods. This survey article addresses these challenges with two primary objectives. First, it systematically reviews and categorizes XAI properties, distinguishing them between human-centered (relying on empirical studies involving explainees) or functionally-grounded (quantitative metrics independent of explainees). Second, it expands this analysis by introducing a hierarchically structured, functionally grounded benchmark framework for XAI methods, providing formal definitions of XAI properties. The framework’s practicality is demonstrated by applying it to two widely used methods: LIME and SHAP.

Graph neural networks have helped researchers overcome the challenges of deep learning on graphs in non-Euclidean space. Like most deep learning algorithms, although the prediction of the models produces good results, explaining the predictions of the model is often challenging. This paper will focus on applying graph neural networks to predict the properties of the various molecules in the molecular datasets. The aim is to explore the generation of explanations for molecule property predictions. The four graph neural networks and seven explainers are chosen to generate and compare the quality of the explanations that are given by the explainers for each of the model predictions. The quality of this explanation is measured by sparsity, fidelity, and fidelity inverse. It is observed that various models find it difficult to learn the node embeddings when there is a class imbalance; despite the models achieving a 75% accuracy and the F1_Score was 66%. It is also observed that for all datasets, sparsity had a statistically significant effect on fidelity; that is, as more important features are masked, the quality of the explanation reduces. The effect of sparsity on fidelity inverse varied from dataset to dataset; as more unimportant features were masked, the quality of the explanations improved in some datasets, yet the change was not significant in other datasets. Finally, it was observed that the explanation quality differs across models. However, larger neural networks produced better predictions in our experiments, and the quality of the explanation of those predictions was not lower than that of smaller neural networks.

Perturbation-based post-hoc image explanation methods are commonly used to explain image prediction models. These methods perturb parts of the input to measure how those parts affect the output. Since the methods only require the input and output, they can be applied to any model, making them a popular choice to explain black-box models. While many different methods exist and have been compared with one another, it remains poorly understood which parameters of the different methods are responsible for their varying performance.

This work uses the Randomized Input Sampling for Explanations (RISE) method as a baseline to evaluate many combinations of mask sampling, segmentation techniques, smoothing, attribution calculation, and per-segment or per-pixel attribution, using a proxy metric. The results show that attribution calculation, which is frequently the focus of other works, has little impact on the results. Conversely, segmentation and per-pixel attribution, rarely examined parameters, have a significant impact.

The availability of easy-to-use and reliable software implementations is important for allowing researchers in academia and industry to test, assess and take into use eXplainable AI (XAI) methods. This paper describes the py-ciu Python implementation of the Contextual Importance and Utility (CIU) model-agnostic, post-hoc explanation method and illustrates capabilities of CIU that go beyond the current state-of-the-art that could be useful for XAI practitioners in general.

Recent advancements in machine learning, particularly in the field of adversarial training, have underscored the necessity for models that are not only accurate but also robust against adversarial attacks. However, a critical challenge lies in the unequal vulnerability of different classes within these models, often leading to a 'weakest-link' phenomenon where certain classes are more susceptible to adversarial perturbations. This paper presents a novel approach to improving the robustness of machine learning models against adversarial attacks, focusing on classes identified as vulnerable. The core of our methodology involves generating adversarial examples specifically tailored to these vulnerable classes, a process we term class-specific perturbations. We integrate this adversarial training into the learning process, aiming to enhance the model's robustness specifically for these at-risk classes. Motivated by this finding, we propose Balanced Adversarial Training (BAT) to facilitate adversarial training for the vulnerable class. Experimental evaluations across different datasets show that our proposed framework not only improves the robustness of the vulnerable class against adversarial attacks but also maintains overall model performance. Our work highlights the importance of moving beyond average accuracy, which is particularly important in safety-critical applications.

 This volume constitutes the papers of several workshops which were held in conjunction with the 6th International Workshop on Explainable and Transparent AI and Multi-Agent Systems, EXTRAAMAS 2024, in Auckland, New Zealand, during May 6–10, 2024.

The 13 full papers presented in this book were carefully reviewed and selected from 25 submissions. The papers are organized in the following topical sections: User-centric XAI; XAI and Reinforcement Learning; Neuro-symbolic AI and Explainable Machine Learning; and XAI & Ethics.

The United Nations launched sustainable development goals in 2015 that include goals for sustainable energy. From global energy consumption, households consume 20–30% of energy in Europe, North America and Asia; furthermore, the overall global energy consumption has steadily increased in the recent decades. Consequently, to meet the increased energy demand and to promote efficient energy consumption, there is a persistent need to develop applications enhancing utilization of energy in buildings. However, despite the potential significance of AI in this area, few surveys have systematically categorized these applications. Therefore, this paper presents a systematic review of the literature, and then creates a novel taxonomy for applications of smart building energy utilization. The contributions of this paper are (a) a systematic review of applications and machine learning methods for smart building energy utilization, (b) a novel taxonomy for the applications, (c) detailed analysis of these solutions and techniques used for the applications (electric grid, smart building energy management and control, maintenance and security, and personalization), and, finally, (d) a discussion on open issues and developments in the field.