Maria Vakalopoulou

Weakly supervised attention maps routinely latch onto spurious confounders—until you regularize them with the minimal attention shift required to flip the model's decision.

Combining hard architectural sparsity with soft regularization can significantly enhance the interpretability of vision models without sacrificing performance.

Tile-level benchmarking can reliably shortlist top-performing models, potentially saving researchers significant time and resources in digital pathology.

Medical vision-language models perform better when the modality gap is tuned to an intermediate level, challenging the assumption that minimizing it is always optimal.