Francesca Toni

Professor Francesca Toni is a Professor in Computational Logic at the Department of Computing, Faculty of Engineering at Imperial College London. She leads research in Artificial Intelligence, focusing on explainable AI (XAI), argumentation theory, and neuro-symbolic systems. Her affiliations include the Centre for eXplainable AI (XAI), Argumentation-based Deep Interactive eXplanations (ADIX), and the Human-Like Computing initiative. Her work integrates computational logic with machine learning to develop interpretable models for healthcare, robotics, and decision support systems. Recent research emphasizes conflict analysis in neural networks, argumentative ensembling, and object-centric learning frameworks.

Her research interests span AI ethics, formal argumentation, and the integration of symbolic reasoning with deep learning. Key contributions include neuro-argumentative learning architectures, benchmarking explainability methods (XAI-Units), and frameworks for robust recourse in model multiplicity scenarios. She actively explores applications in biomedical fraud detection (Pub-Guard-LLM) and personalized decision support via gradual bipolar argumentation.

Her publications highlight trends in explainable AI, with a focus on visual debates, counterfactual explanations, and causal structure learning. She has pioneered systems like ProtoArgNet for interpretable image classification and DR-HAI for dialectical reconciliation in human-AI interactions. Her work bridges theoretical foundations (e.g., ABA semantics) with real-world applications in healthcare and legal reasoning.

Notable projects include ROAD2H—an open-source XAI approach for managing comorbidities—and Cafe for conflict-aware feature explanations. She has contributed to legal AI systems (LawGIBA) and causal discovery methods. Current efforts focus on neuro-argumentative machine learning and object-centric representation learning.