Publications

Working papers

When Imbalance Comes Twice: Active Learning under Simulated Class Imbalance and Label Shift in Binary Semantic Segmentation

This preprint was released in January 2026 and was motivated by the fact that machine vision is prone to strong label shift, and we were worried that adding uncertainty based active learning strategy would increase the distance between the production and database distributions, thus worsening models predictions. The contribution is humble but it was a fun project that provided insights about active learning in industry.

Publications

Hope it will not be empty at the end of the Phd 😢.