Transfer learning for honey bee toxicity prediction: MolFormer versus classical QSAR representations
Alan Victor de Souza Pinho, Edilson Beserra de Alencar Filho, Rosalvo Ferreira de Oliveira Neto · 2026
Read Transfer learning for honey bee toxicity prediction: MolFormer versus classical QSAR representations by Alan Victor de Souza Pinho, Edilson Beserra de Alencar Filho, Rosalvo Ferreira de Oliveira Neto (2026) 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-13534182.txt.
Description
Pinho, Alencar Filho, and Oliveira Neto compare models that predict harm to bees, in work from twenty twenty-six. Testing farm chemicals on live bees is costly and data are scarce. Computer forecasts could screen chemicals first, if they are accurate enough.
The dataset held about a thousand compounds, classed as toxic or safe for bees. Chemical forms were coded in three ways: classic descriptors, fingerprints, and embeddings from a pretrained language style model. Three standard classifiers were trained on each form. The study asks whether transfer learning from a large chemical model beats classic coding for this small, vital safety task.
Catalog record
| CREATOR | Alan Victor de Souza Pinho, Edilson Beserra de Alencar Filho, Rosalvo Ferreira de Oliveira Neto |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |