Benchmarking MIL methods

This project is part of the CS‑559 “Machine Learning” course at EPFL, taught by Prof. Martin Jaggi and Nicolas Flammarion, and the “ML4Science” program, enabling students to work on real-world scientific problems in collaboration with research labs.

Conducted with Prof. Charlotte Bunne’s lab in Artificial Intelligence in Molecular Medicine, the project benchmarks Multiple Instance Learning (MIL) methods for classifying Whole Slide Images (WSI) in histopathology. Various MIL algorithms are implemented and evaluated to identify the most effective approaches for handling large, complex tissue images.

You can find below the project report and code repository.