This paper provides new theory to support to the eXplainable AI (XAI) method Contextual Importance and Utility (CIU). CIU arithmetic is based on the concepts of Multi-Attribute Utility Theory, which gives CIU a solid theoretical foundation. The novel concept of contextual influence is also defined, which makes it possible to compare CIU directly with so-called additive feature attribution (AFA) methods for model-agnostic outcome explanation. One key takeaway is that the "influence" concept used by AFA methods is inadequate for outcome explanation purposes even for simple models to explain. Experiments with simple models show that explanations using contextual importance (CI) and contextual utility (CU) produce explanations where influence-based methods fail. It is also shown that CI and CU guarantees explanation faithfulness towards the explained model.

Many existing methods of counterfactual explanations ignore the intrinsic relationships between data attributes and thus fail to generate realistic counterfactuals. Moreover, the existing methods that account for relationships between data attributes require domain knowledge, which limits their applicability in complex real-world applications. In this paper, we propose a novel approach to realistic counterfactual explanations that preserve relationships between data attributes. The model directly learns the relationships by a variational auto-encoder without domain knowledge and then learns to disturb the latent space accordingly. We conduct extensive experiments on both synthetic and real-world datasets. The results demonstrate that the proposed method learns relationships from the data and preserves these relationships in generated counterfactuals.

Although standard Machine Learning models are optimized for making predictions about observations, more and more they are used for making predictions about the results of actions. An important goal of Explainable Artificial Intelligence (XAI) is to compensate for this mismatch by offering explanations about the predictions of an ML-model which ensure that they are reliably action-guiding. As action-guiding explanations are causal explanations, the literature on this topic is starting to embrace insights from the literature on causal models. Here I take a step further down this path by formally defining the causal notions of sufficient explanations and counterfactual explanations. I show how these notions relate to (and improve upon) existing work, and motivate their adequacy by illustrating how different explanations are action-guiding under different circumstances. Moreover, this work is the first to offer a formal definition of actual causation that is founded entirely in action-guiding explanations. Although the definitions are motivated by a focus on XAI, the analysis of causal explanation and actual causation applies in general. I also touch upon the significance of this work for fairness in AI by showing how actual causation can be used to improve the idea of path-specific counterfactual fairness.

This paper deals with the problem of finding the preferred extensions of an argumentation framework by means of a bijection with the naive extensions of another framework. First we consider the case where an argumentation framework is naive-realizable: its naive and preferred extensions are equal. Recognizing naive-realizable argumentation frameworks is hard, but we show that it is tractable for frameworks with bounded in-degree. Next, we give a bijection between the preferred extensions of an argumentation framework being admissible-closed (the intersection of two admissible sets is admissible) and the naive extensions of another framework on the same set of arguments. On the other hand, we prove that identifying admissible-closed argumentation frameworks is coNP-complete. At last, we introduce the notion of irreducible self-defending sets as those that are not the union of others. It turns out there exists a bijection between the preferred extensions of an argumentation framework and the naive extensions of a framework on its irreducible self-defending sets. Consequently, the preferred extensions of argumentation frameworks with some lattice properties can be listed with polynomial delay and polynomial space.

In the future, AI will increasingly find its way into systems that can potentially cause physical harm to humans. For such safety-critical systems, it must be demonstrated that their residual risk does not exceed what is acceptable. This includes, in particular, the AI components that are part of such systems' safety-related functions. Assurance cases are an intensively discussed option today for specifying a sound and comprehensive safety argument to demonstrate a system's safety. In previous work, it has been suggested to argue safety for AI components by structuring assurance cases based on two complementary risk acceptance criteria. One of these criteria is used to derive quantitative targets regarding the AI. The argumentation structures commonly proposed to show the achievement of such quantitative targets, however, focus on failure rates from statistical testing. Further important aspects are only considered in a qualitative manner -- if at all. In contrast, this paper proposes a more holistic argumentation structure for having achieved the target, namely a structure that integrates test results with runtime aspects and the impact of scope compliance and test data quality in a quantitative manner. We elaborate different argumentation options, present the underlying mathematical considerations, and discuss resulting implications for their practical application. Using the proposed argumentation structure might not only increase the integrity of assurance cases but may also allow claims on quantitative targets that would not be justifiable otherwise.

Recently, (Dunne et al. 2009; 2011) have suggested to weight attacks within Dung's abstract argumentation frameworks, and introduced the concept of WAF (Weighted Argumentation Framework). However, they use WAFs in a very specific way for relaxing attacks. The aim of this paper is to explore ways to take advantage of attacks weights within an argumentation process. Two different approaches are considered: The first one extends the proposal by (Dunne et al. 2011) and accounts for other aggregation functions than sum in the objective of relaxing attacks. The second one shows how weights can be exploited to strengthen the usual notion of defence, leading to new concepts of extensions.

Argumentation is a human-like reasoning mechanism contributing to the formalization of commonsense reasoning. In the last decade, several argument-based formalisms have emerged, with application in many areas, such as legal reasoning, autonomous agents and multi-agent systems; many are based on Dung's seminal work characterizing Abstract Argumentation Frameworks (AF). Recent research in the area has led to Temporal Argumentation Frameworks (TAF) that extend Dung's by considering the temporal availability of arguments. In this work we introduce a novel framework, called Extended Temporal Argumentation Framework (E-TAF), extending TAF with the capability of modeling availability of attacks among arguments, which allows for instance to model reliability of arguments varying over time. We show how E-TAF can be enriched by considering Structured Abstract Argumentation, adding compositional elements to the abstract arguments involved based on a simplified version of the recently introduced Dynamic Argumentation Frameworks.

Fixpoints play a key role in the mathematical set up of abstract argumentation theory but, we argue, have been relatively underexamined in the literature. The paper studies the logical structure underlying the computation via approximation sequences of the sort of fixpoints relevant in argumentation. Concretely, it presents a number of novel results on the fixed point theory underpinning the main Dung's semantics and, inspired by recent literature on the logical analysis of equilibrium computation in games, it provides a characterization of those semantics in terms of iterated model updates.

argumentation frameworks (AFs) provide the basis for various reasoning problems in the areas of Knowledge Representation and Artificial Intelligence. Efficient evaluation of AFs has thus been identified as an important research challenge. So far, implemented systems for evaluating AFs have either followed a straight-forward reduction-based approach or been limited to certain tractable classes of AFs. In this work, we present a generic approach for reasoning over AFs, based on the novel concept of complexity-sensitivity. Establishing the theoretical foundations of this approach, we derive several new complexity results for preferred, semistable and stage semantics which complement the current complexity landscape for abstract argumentation, providing further understanding on the sources of intractability of AF reasoning problems. The introduced generic framework exploits decision procedures for problems of lower complexity whenever possible. This allows, in particular, instantiations of the generic framework via harnessing in an iterative way current sophisticated Boolean satisfiability (SAT) solver technology for solving the considered AF reasoning problems. First experimental results show that the SAT-based instantiation of our novel approach outperforms existing systems.

We contribute to the investigation of possible outcomes of argumentation under semantics formulated using argumentation frameworks (AFs). In particular, we study this question for the labelling-based formulation of such semantics, generalizing previous work which has focused on extensions. In this paper, we restrict attention to the preferred and semi-stable semantics, showing that as long as we have a sufficient number of fresh arguments available, we can in fact argue for anything. That is, for any set of finite labellings there is an AF that enforces exactly this set as the outcome of argumentation.