Get Started
Home
Topics
Search
Library
Research questionHow can pretrained transformer embeddings capture peptide properties from scarce labels without distorting their geometry?Scarce labeled peptide data make it difficult to extract task-relevant signals from pretrained embeddings. Adaptation can either reshape the original representation substantially or fail to capture properties needed for downstream design tasks.
AI
Diffusion Models
Health
Machine Learning
Research Paper
Latest papersRecent research connected to this question, newest first.Freeze, Diffuse, Decode: Task-Aware Adaptation of Transformer Embeddings for Antimicrobial Peptide DesignThe setting is antimicrobial peptide design with pretrained transformer embeddings and limited supervised data. The reported downstream uses include property prediction, retrieval, and latent-space interpolation, with an emphasis on low-dimensional and interpretable representations.research paper · Sep 2, 2026
Related questions
How can self-supervised representations transfer reliably across microscopy datasets with scarce labels and mismatched staining or channels?How can molecular property predictors retain substructure and graph-distance information in compact representations without external pretraining?How should pretrained image representations be chosen to cluster unseen datasets without labels or retraining?How can protein design models jointly model amino acid sequences and 3D structures without separate autoencoder pretraining?