This workshop aims to spearhead research on Human-Interpretable Artificial Intelligence (HI-AI) by providing: (i) a general overview of the key aspects of HI-AI, in order to equip all researchers with the necessary background and set of definitions; (ii) novel and interesting ideas coming from both invited talks and top paper contributions; (iii) the chance to engage in dialogue with prominent scientists during poster presentations and cofee breaks. The workshop welcomes contributions covering novel interpretableby-design or post-hoc approaches, as well as theoretical analysis of existing works. Additionally, we accept visionary contributions speculating on the future potential of this field. Finally, we welcome contributions from related fields such as Ethical AI, Knowledgedriven Machine learning, Human-machine Interaction, applications in Medicine and Industry, and analyses from Regulatory experts. • Computing methodologies → Artificial intelligence .
Human-interpretable AI models [
To really explain DNN decision-making process, there is a
growing consensus that human-interpretable explanations are required.
Human-Interpretable AI (HI-AI) methods either provide post-hoc
explanations by extracting the symbols that have been
automatically learnt by the models (e.g., T-CAV [
Topics of interest include, but are not limited to, the following: • Explainable-by-design models, novel approaches to creating machine learning and deep learning models that are intrinsically explainable or interpretable. • Post-hoc methods for Interpretable AI, novel approaches on post-hoc interpretable AI. These include but are not limited to approaches working on higher-level features such as concepts. • Theoretical analyses of existing methods, showing what existing interpretable methods can achieve both from an explanation and a generalization point of view. • Knowledge integration & Reasoning methods injecting domain knowledge and reasoning methods into deep learning models to enhance their interpretability and performance. • AI Ethics papers analysing implications of interpretable AI methods, discussing topics such as fairness, accountability, transparency, and bias mitigation in AI systems. • Human-machine Interaction studies on innovative humanmachine interaction systems, successfully exploiting interpretable AI models in their capability to provide both standard and counter-factual explanations. • Vision papers on XAI discussing the possible evolutions of the XAI field or speculating potential interpretable system and applications with their implications. • Applications in Medicine and Healthcare applications of interpretable AI methods in medical diagnosis, treatment planning, and healthcare decision-making. • AI in Industry practical applications of interpretable AI methods in various safety-critical industrial sectors, such as transportation, finance and retail. • Legal and Regulatory dissertations discussing and providing analysis of the legal challenges associated with interpretable AI, including compliance with data protection laws for transparent and accountable AI systems.
Program Outline. Table 1 reports the workshop program. Firstly, we will give an overview of the key aspects of HI-AI to ensure all attendees have a solid understanding of the background concepts and terminology. Secondly, the workshop features three invited talks from experts in the field, who will share their insights and latest research findings. These talks will provide valuable perspectives and inspire new ideas. Thirdly, we will ofer participants the chance to engage in dialogue with prominent scientists during a long cofee break with poster presentations, encouraging collaborations and knowledge-sharing. Also, the workshop program includes three contributed talks from selected contributions. We will recognize the most interesting contribution with a Best Workshop Paper Award. We have allocated 40 minutes for each invited talk, allowing for a 30-minute presentation followed by a 10-minute Q&A session. We allotted the same time for the poster sessions. 4
Paper management. We published the Call For Papers (CFP) on the workshop website1. The CFP focuses on short papers, which can be research papers, theoretical analysis papers, or vision papers.
In the case of research contributions, we asked paper authors to make their code and data openly available to ensure reproducibility. The review process has been double-blind. We have used OpenReview to ensure the final decisions for each paper are made by the organisers with no conflict of interest. All accepted papers will be published on the workshop website, which will remain active and accessible after the conference concludes. Additionally, we took contact with an external editor (CEUR-WS) to create an archival version of these papers for authors who wish to participate in a subsequent publication.
We are very grateful to each of our program committee members for their hard reviewing work, namely Romain Giot, Eliana Pastor, Roberto Pellungrini, Eleonora Poeta, Gianluigi Lopardo, and Gizem Gezici, besides workshop chairs.