LoRA boosts SAS vision transformer AUPRC to 0.679 in underwater target recognition
Researchers adapted a DINOv3 Vision Transformer (ViT) for synthetic aperture sonar (SAS) automatic target recognition, a task limited by scarce imagery and noisy acoustic backgrounds. Their three‑stage, parameter‑efficient framework first applies Low‑Rank Adaptation (LoRA) while keeping the ViT backbone frozen, then adds hard‑negative mining, and finally uses supervised contrastive learning…
1 source primary source
Key points
- LoRA adaptation raised AUPRC from 0.300 to 0.679 ± 0.027 using a frozen ViT backbone.
- Only 0.26 % of model weights (Rank 4) were trained during LoRA adaptation.
- Hard‑negative mining and SupCon produced negligible AUPRC changes (‑0.0045 ± 0.0119 and +0.0002 ± 0.0096).
Read the original atarXiv cs.AI · by Dan Zimmerman, Frank E. Bobe III, Amelia L. McCormack, Matthew Cook, Gregory D. Vetaw primary sourceOpen source ↗
Topics · follow one to build your own front page
DINOv3 Vision Transformer
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.
More in Research
All →- TatBLiMP benchmarks Tatar linguistic minimal pairs · 1 src
- Recursive language models generalize out of domain, study shows · 1 src
- reviser proposes cursor-based text generation · 1 src
- SAGE system raises grant review agreement to kappa 0.58, beating baseline · 1 src
- Qwen2.5-Omni-3B adapters boost entity recall in accented conversational ASR · 1 src
Comments
via GitHub Discussions