Explainable AI xAI 2021

TOPICS of this workshop:

In-line with the general theme of the CD-MAKE conference of augmenting human intelligence with artificial intelligence, and Science is to test crazy ideas –  Engineering is to bring these ideas into Business – we foster cross-disciplinary and interdisciplinary work in order to bring together experts from different fields, e.g. computer science, psychology, sociology, philosophy, law, business, … experts who would otherwise possibly not meet together. This cross-domain integration and appraisal of different fields of science and industry shall provide an atmosphere to foster different perspectives and opinions; it will offer a platform for novel crazy ideas and a fresh look on the methodologies to put these ideas into business.

Topics include but are not limited to (alphabetically – not prioritized):

• Abstraction of human explanations (“How can xAI learn from human explanation straregies?”)
• Acceptance (“How to ensure acceptance of AI/ML among end users?”)
• Accountability and responsibility (“Who is to blame if something goes wrong?” “What legal aspects does it involve?”)
• Action Influence Graphs (“How can they help interpretability?”)
• Active machine learning algorithm design for interpretability purposes
• Adversarial attacts detection, explanation and defense (“How can we interpret adversarial examples?”)
• Adaptive personal xAI systems (“To whom, when, how to provide explanations?”)
• Adaptable explanation interfaces (“How can we adapt xAI-interfaces to the needs, demands, requirements of end-users and domain experts?”)
• Affective computing for successful human-AI interaction  and human-robot interaction
• Application of xAI methods (e.g. in smart health, smart farming, forest ecosystems, …)
• Argumentation theories of explanations
• Artificial advice givers
• Bayesian rule lists
• Bayesian modeling and optimization (“How to design efficient methods for learning interpretable models?”)
• Bias and fairness in explainable AI (“How to avoid bias in machin learning applications?”)
• Bridiging the gap between humans and machines (concepts, methods, tools, …)
• Causal learning, causal discovery, causal reasoning, causal explanations, and causal inference
• Causality research in the sense of Judea Pearl
• Causability research (“measuring the Quality of explanations“)
• Cognitive issues of explanation and understanding (“understanding understanding”)
• Combination of statistical learning approaches with large knowledge repositories (ontologies, terminologies, …)
• Combination of deep learning approaches with traditional AI approaches
• Comparison of human intelligence vs. artificial intelligence (HCI — KDD)
• Computational behavioural science (“How are people thinking, judging, making decisions – and explaining it?”)
• Constraints-based explanations
• Counterfactual explanations (“What did not happen?”, “How can we provide counterfactual explanations?”)
• Contrastive explanation methods (CEM) as e.g. in Criminology or medical diagnosis
• Cyber security, cyber defense and malicious use of adversarial examples
• Data dredging explainability and causal inference
• Decision making and decision support systems (“Is a human-like decision good enough?)
• Dialogue systems for enhanced human-ai interaction
• Emotional intelligence (“Emotion AI”) and emotional UI
• Ethical aspects of AI in general and human-AI interaction in particular
• Evaluation criteria
• Explanation agents and recommender systems
• Explanatory user interfaces and Human-Computer Interaction (HCI) for explainable AI
• Explainable reinforcement learning
• Explainable and verifiable activity recognition
• Explaining agent behaviour (“How to know if the agent is going to make a mistake and when?”)
• Explaining robot behviour (“Why did you take action x in state s?”)
• Fairness, accountability and trust (“How to ensure trust in AI?”)
• Frameworks, architectures, algorithms and tools to support post-hoc and ante-hoc explainability
• Frameworks for reasoning about causality
• Gradient based interpretability to understand data sensitivity
• Graphical causal inference and graphical models for explanation and causality
• Ground truth
• Group recommender systems
• Human-AI interaction and intelligent interfaces
• Human-AI teaming for ensuring trustworthy AI systems
• Human-centered AI
• Human-in-the-loop learning approaches, methodologies, tools and systems
• Human rights vs. robot rights
• Implicit knowlege elicitation
• Industrial applications of xAI, e.g. in medicine, autonomous driving, production, finance, ambient assisted living, etc.
• Integrating of deep learning approaches with grammars of graphical models
• Interactive machine learning with a human-in-the-loop
• Interactive machine learning with (many) humans-in-the-loop (crowd intelligence)
• Interpretability in ranking algorithms (“How to explain ranking algorithms, e.g. patient ranking in helath, in human interpretable ways?”)
• Interpretability in reinforement learning
• Interpretable representation learning (“How to make sense of data assigned to similar representations?”)
• Kandinsky Patterns experiments and extensions
• Legal aspects of AI/ML (“Who is to blame if an error occurs?”)
• Metrics for evaluation of the quality of explanations
• Misinformation in Social Media, Ground truth evaluation and explanation
• Model explainability, quality and provenance
• Moral principles and moral dilemmas of current and future AI
• Natural Language Argumentation interfaces for explanation generation
• Natural Language generation for explanatory models
• Non-IID learning models, algorithms, analytics, recommenders
• Novel intelligent future user interfaces (e.g. affective mobile interfaces)
• Novel methods, algorithms, tools, procedures for supporting explainability in the AI/ML pipeline
• Personalized xAI
• Philosophical approaches of explainability, theories of mind (“When is it enough explained? Do we have a degree of saturation?”)
• Policy explanations to humans (“How to explain and why is the next step the best action to select?”)
• Proof-of-concepts and demonstrators of how to integrate explainable AI into real-world workflows and industrial processes
• Privacy, surveillance, control and agency
• Psychology of human concept learning and tranfer to machine learning
• Python for nerds (Python tricks of the trade – relevant for explainable AI)
• Quality of explanations and how to measure quality of explanations
• Real-World success stories of xAI
• Rendering of reasoning processes
• Reproducibility, replicability, retraceability, reenactivity
• Self-explanatory agents and decision support systems
• Similarity measures for xAI
• Social implications of AI (“What AI impacts”), e.g. labour trends, human-human interaction, machine-machine interaction
• Soft Decision Trees (SDT)
• Spartanic approaches of explanations (“What is the most simplest explanation?”)
• Structureal causal equations (SCM)
• Theoretical approaches of explainability (“What makes a good explanation?”)
• Theories of explainable/interpretable models
• Tools for model understanding (diagnostic, debugging, introspection, visualization, …)
• Transparent reasoning
• Trustworthy human-AI teaming under uncertainties
• Understanding understanding
• Understanding Markov decision processes and prtially observable markov decision processes
• Usability of Human-AI interfaces
• Visualizing learned representations
• Web- and mobile-based cooperative intelligent information systems and tools