Transfer Learning Approach with Features Block Selection via Genetic Algorithm for High-Imbalance and Multi-Label Classification of HPA Confocal Microscopy Images
Vincenzo Taormina, Domenico Tegolo, Cesare Valenti, Fabiano Bini · 2025
Read Transfer Learning Approach with Features Block Selection via Genetic Algorithm for High-Imbalance and Multi-Label Classification of HPA Confocal Microscopy Images by Vincenzo Taormina, Domenico Tegolo, Cesare Valenti, Fabiano Bini (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12730114.txt.
Description
Vincenzo Taormina and co-authors published this study in 2025. It classifies protein locations in microscope images from the Human Protein Atlas. The task is hard: twenty-eight location patterns, heavy imbalance, and cells that carry several labels at once. Their approach borrows features from twelve networks pre-trained on everyday photos. A genetic algorithm then picks the best mix of feature blocks to keep.
Simple joining of blocks already beat single networks. The tuned mix did better still. The team also split analysis between whole images and single cells, reading the nucleus and its rim separately. That cell-level view, new for this data set, caught rare patterns well. One rare class scored 0.8 on the F1 measure, topping past work. Channel tests showed the protein channel strongest alone, but richer in combination.
Catalog record
| CREATOR | Vincenzo Taormina, Domenico Tegolo, Cesare Valenti, Fabiano Bini |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 84 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |