Andrew Cropper

Andrew Cropper is a Research Fellow in the Department of Computer Science at the University of Oxford. His research focuses on the intersection of logic and machine learning, particularly inductive logic programming (ILP) and program synthesis. He leads the Logic and Learning research group and the 'The Automatic Computer Scientist' project, which explores automated program generation and symbolic learning methods. His work emphasizes developing efficient algorithms for learning logical programs and improving the scalability of inductive logic programming.

Recent publications highlight advancements in combining small logical rules into larger structures, discovering higher-order abstractions in logic programs, and applying symbolic metaprogramming to enhance learning efficiency. His research also addresses challenges in handling noisy data and integrating numerical reasoning into relational program synthesis. Collaborations with cognitive scientists explore parallels between human rule-learning processes and machine learning algorithms.

Cropper has advised students including Rolf Morel (past researcher) and collaborates with researchers like Céline Hocquette and Matti Järvisalo. His contributions span over 30 peer-reviewed articles in top conferences (IJCAI, AAAI) and journals (Nature Communications, Machine Learning). He maintains active roles in the ILP community, co-authoring comprehensive reviews of the field's 30-year evolution and advancing its applications in program induction and general game playing.

Research highlights include developing the Metagol system for ILP, pioneering predicate invention techniques, and introducing constraint-driven methods to improve program synthesis efficiency. His work bridges symbolic AI with modern machine learning, aiming to create interpretable and generalizable algorithms for complex problem-solving.