LASIGE Talks are fortnightly/monthly events to publicize recently distinguished publications or ongoing cutting-edge work by researchers from the research centre, consolidating the scientific culture of the LASIGE community.

Speakers: Alcides Fonseca and Hugo Figueiras
Date: March 25th, 2026, Wednesday, 12:00
Where: C6.3.27

12:00 Talk by Alcides Fonseca
12:20 Talk by Hugo Figueiras
12:40 Q&A + Break for snacks & coffee

Talk1: Safer Software with Liquid Types
Speaker: Alcides Fonseca
Abstract: In a world where LLM-generated code is being produced at a faster pace than human written code, verification is more important than ever. Liquid Types (refining types with logical predicates, e.g. {x:Int | x > 10}) have been around for 17 years now but, despite their many applications, they haven’t taken off. In this talk we will answer why (PLDI’25), based on user interviews we conducted, relating them to other verification tools such as Interactive Theorem Provers and Design-by-Contract approaches like Dafny. Finally, we will see how our research group is addressing those challenges in both LiquidJava and Aeon.
Suggested reading: https://dl.acm.org/doi/pdf/10.1145/3729327

Talk2: Self-supervised learning for breast cancer detection: A review
Speaker: Hugo Figueiras
Abstract: Breast cancer remains a leading cause of mortality, driving advances in detection and diagnosis. Deep learning aids CAD systems but depends on costly labeled data. Self-supervised learning (SSL) addresses this by leveraging unlabeled images to learn robust representations. This review examines SSL across screening, diagnosis, grading, and staging, covering mammography, DBT, ultrasound, MRI, and histopathology. SSL reduces annotation needs, improves generalization, and enhances localization. While successful in several modalities, PET remains underexplored. We highlight future directions, including multimodal fusion, domain-adaptive tasks, and explainable models, emphasizing SSL’s potential for scalable, clinically useful breast cancer imaging.
Paper: https://doi.org/10.1016/j.compbiomed.2025.111245