The Workshop Program of the Association for the Advancement of Artificial Intelligence's Thirty-Fifth Conference on Artificial Intelligence was held virtually from February 8-9, 2021. There were twenty-six workshops in the program: Affective Content Analysis, AI for Behavior Change, AI for Urban Mobility, Artificial Intelligence Safety, Combating Online Hostile Posts in Regional Languages during Emergency Situations, Commonsense Knowledge Graphs, Content Authoring and Design, Deep Learning on Graphs: Methods and Applications, Designing AI for Telehealth, 9th Dialog System Technology Challenge, Explainable Agency in Artificial Intelligence, Graphs and More Complex Structures for Learning and Reasoning, 5th International Workshop on Health Intelligence, Hybrid Artificial Intelligence, Imagining Post-COVID Education with AI, Knowledge Discovery from Unstructured Data in Financial Services, Learning Network Architecture During Training, Meta-Learning and Co-Hosted Competition, ...

Insurance is moving toward a world in which carriers will not be allowed to make decisions that affect customers based on black-box AI. Artificial intelligence (AI) holds a lot of promise for the insurance industry, particularly for reducing premium leakage, accelerating claims and making underwriting more accurate. AI can identify patterns and indicators of risk that would otherwise go unnoticed by human eyes. Unfortunately, AI has often been a black box: Data goes in, results come out and no one -- not even the creators of the AI -- has any idea how the AI came to its conclusions. That's because pure machine learning (ML) analyzes the data in an iterative fashion to develop a model, and that process is simply not available or understandable.

The pyglaf reasoner takes advantage of circumscription to solve computational problems of abstract argumentation frameworks. In fact, many of these problems are reduced to circumscription by means of linear encodings, and a few others are solved by means of a sequence of calls to an oracle for circumscription. Within pyglaf, Python is used to build the encodings and to control the execution of the external circumscription solver, which extends the SAT solver glucose and implements algorithms taking advantage of unsatisfiable core analysis and incremental computation.

We present Fudge, an abstract argumentation solver that tightly integrates satisfiability solving technology to solve a series of abstract argumentation problems. While most of the encodings used by Fudge derive from standard translation approaches, Fudge makes use of completely novel encodings to solve the skeptical reasoning problem wrt. preferred semantics and problems wrt. ideal semantics.

We propose CX-ToM, short for counterfactual explanations with theory-of mind, a new explainable AI (XAI) framework for explaining decisions made by a deep convolutional neural network (CNN). In contrast to the current methods in XAI that generate explanations as a single shot response, we pose explanation as an iterative communication process, i.e. dialog, between the machine and human user. More concretely, our CX-ToM framework generates sequence of explanations in a dialog by mediating the differences between the minds of machine and human user. To do this, we use Theory of Mind (ToM) which helps us in explicitly modeling human's intention, machine's mind as inferred by the human as well as human's mind as inferred by the machine. Moreover, most state-of-the-art XAI frameworks provide attention (or heat map) based explanations. In our work, we show that these attention based explanations are not sufficient for increasing human trust in the underlying CNN model. In CX-ToM, we instead use counterfactual explanations called fault-lines which we define as follows: given an input image I for which a CNN classification model M predicts class c_pred, a fault-line identifies the minimal semantic-level features (e.g., stripes on zebra, pointed ears of dog), referred to as explainable concepts, that need to be added to or deleted from I in order to alter the classification category of I by M to another specified class c_alt. We argue that, due to the iterative, conceptual and counterfactual nature of CX-ToM explanations, our framework is practical and more natural for both expert and non-expert users to understand the internal workings of complex deep learning models. Extensive quantitative and qualitative experiments verify our hypotheses, demonstrating that our CX-ToM significantly outperforms the state-of-the-art explainable AI models.

Established approaches to assuring safety-critical systems and software are difficult to apply to systems employing machine learning (ML). In many cases, ML is used on ill-defined problems, e.g. optimising sepsis treatment, where there is no clear, pre-defined specification against which to assess validity. This problem is exacerbated by the "opaque" nature of ML where the learnt model is not amenable to human scrutiny. Explainable AI methods have been proposed to tackle this issue by producing human-interpretable representations of ML models which can help users to gain confidence and build trust in the ML system. However, there is not much work explicitly investigating the role of explainability for safety assurance in the context of ML development. This paper identifies ways in which explainable AI methods can contribute to safety assurance of ML-based systems. It then uses a concrete ML-based clinical decision support system, concerning weaning of patients from mechanical ventilation, to demonstrate how explainable AI methods can be employed to produce evidence to support safety assurance. The results are also represented in a safety argument to show where, and in what way, explainable AI methods can contribute to a safety case. Overall, we conclude that explainable AI methods have a valuable role in safety assurance of ML-based systems in healthcare but that they are not sufficient in themselves to assure safety.

Abstract argumentation provides us with methods such as gradual and Dung semantics with which to evaluate arguments after potential attacks by other arguments. Some of these methods can take intrinsic strengths of arguments as input, with which to modulate the effects of attacks between arguments. Coming from abstract argumentation, these methods look only at the relations between arguments and not at the structure of the arguments themselves. In structured argumentation the way an argument is constructed, by chaining inference rules starting from premises, is taken into consideration. In this paper we study methods for assigning an argument its intrinsic strength, based on the strengths of the premises and inference rules used to form said argument. We first define a set of principles, which are properties that strength assigning methods might satisfy. We then propose two such methods and analyse which principles they satisfy. Finally, we present a generalised system for creating novel strength assigning methods and speak to the properties of this system regarding the proposed principles.

One of the biggest challenges of machine learning and artificial intelligence is their inability to explain their decision to the users. This black box in AI renders the system largely impenetrable, making it difficult for scientists and researchers to understand why a certain system is behaving the way it is. In recent years, a new branch of explainable AI (XAI) has emerged, which the researchers are actively pursuing to establish user-friendly AI. That said, how AI explanations are perceived is highly dependent on a person's background in AI. A new study named "The Who in Explainable AI: How AI Background Shapes Perceptions of AI Explanations", argues that AI background influences each group's interpretations and that these differences exist through the lens of appropriation and cognitive heuristics.

Natural language free-text explanation generation is an efficient approach to train explainable language processing models for commonsense-knowledge-requiring tasks. The most predominant form of these models is the explain-then-predict (EtP) structure, which first generates explanations and uses them for making decisions. The performance of EtP models is highly dependent on that of the explainer by the nature of their structure. Therefore, large-sized explanation data are required to train a good explainer model. However, annotating explanations is expensive. Also, recent works reveal that free-text explanations might not convey sufficient information for decision making. These facts cast doubts on the effectiveness of EtP models. In this paper, we argue that the predict-then-explain (PtE) architecture is a more efficient approach in terms of the modelling perspective. Our main contribution is twofold. First, we show that the PtE structure is the most data-efficient approach when explanation data are lacking. Second, we reveal that the PtE structure is always more training-efficient than the EtP structure. We also provide experimental results that confirm the theoretical advantages.

Machine Learning (ML)-based network intrusion detection systems bring many benefits for enhancing the cybersecurity posture of an organisation. Many systems have been designed and developed in the research community, often achieving a close to perfect detection rate when evaluated using synthetic datasets. However, the high number of academic research has not often translated into practical deployments. There are several causes contributing towards the wide gap between research and production, such as the limited ability of comprehensive evaluation of ML models and lack of understanding of internal ML operations. This paper tightens the gap by evaluating the generalisability of a common feature set to different network environments and attack scenarios. Therefore, two feature sets (NetFlow and CICFlowMeter) have been evaluated in terms of detection accuracy across three key datasets, i.e., CSE-CIC-IDS2018, BoT-IoT, and ToN-IoT. The results show the superiority of the NetFlow feature set in enhancing the ML models detection accuracy of various network attacks. In addition, due to the complexity of the learning models, SHapley Additive exPlanations (SHAP), an explainable AI methodology, has been adopted to explain and interpret the classification decisions of ML models. The Shapley values of two common feature sets have been analysed across multiple datasets to determine the influence contributed by each feature towards the final ML prediction.