Learning with Logic: Neuro-Symbolic AI for Transparent and Robust Decision-Making by Prof Alessandra Russo
Abstract
AI systems have achieved remarkable capabilities across a wide range of tasks, yet their adoption in high-stakes domains remains limited by fundamental concerns around interpretability, robustness, and reliable generalisation. Addressing these challenges requires methods that integrate data-driven learning with explicit representations of knowledge, structure, and reasoning.
In the first part of this talk, I will present a neuro-symbolic perspective on these challenges, grounded in work on learning from answer sets — a form of symbolic machine learning that produces interpretable models from noisy data in a way that is data-efficient, scalable, and robust. I will explore how symbolic representations and logical semantics can be integrated with neural architectures to support transparent inference and reasoning across a range of settings.
In the second part, I will illustrate these ideas through applications, including policy learning, semantically controlled generation and clinical decision-making. I discuss how neuro-symbolic methods can be applied to learn interpretable models from patient data and support clinicians with recommendations that are both meaningful and auditable in real-world healthcare settings.
The central question unifying this work is how to build AI systems that are not only accurate, but interpretable, robust, and capable of supporting decisions that generalise reliably beyond the conditions they were trained on.
About the Speaker
Alessandra Russo is Professor of Applied Computational Logic at Imperial College London, where she leads the SPIKE research group. She has held major leadership roles, including Head of the Department of Computing (2024–2025), Chair of Imperial-X on interdisciplinary AI, and Convening Co-Director of Human and Artificial Intelligence at the School of Convergence Science. Since 2021, she has co-directed the UKRI Centre for Doctoral Training in Safe and Trusted AI.
Her research focuses on developing interpretable and trustworthy AI systems capable of learning from complex, noisy data, particularly for high-stakes domains like healthcare. She is best known for ILASP, a system that learns transparent, human-readable rules in a robust and auditable way. Her work advances neuro-symbolic AI, combining machine learning with symbolic reasoning to create accurate, safe, and generalisable systems. Professor Russo has led numerous projects, published over 200 papers, and is a sought-after international keynote speaker.