AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy

Yiting Yang, Hao Luo, Yuan Sun, Qingsen Yan, Haokui Zhang, Wei Dong, Guoqing Wang, Peng Wang, Yang Yang, Hengtao Shen

ICCV 2025 · Architectures and efficiency › Pruning and efficient tuning · approach: Efficient ViT adaptation · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace LoRA matrices with AOFTLoRA · VTAB-1k ViT-B/16mean accuracy: 72.3 -> 72.9Improved
replace Adapter matrices with AOFTAdapter · VTAB-1k ViT-B/16mean accuracy: 71.4 -> 72.5Improved
replace Adapter matrices with AOFTAdapter · FGVC ViT-B/16mean accuracy: 85.7 -> 88.9Improved
replace VPT-Shallow prompts with AOFTVPT-Shallow · FGVC ViT-B/16mean accuracy: 84.7 -> 84.0Hurt
replace VPT-Shallow prompts with AOFTVPT-Shallow · VTAB-1k ViT-B/16mean accuracy: 64.8 -> 66.8Improved
replace VPT-Deep prompts with AOFTVPT-Deep · VTAB-1k ViT-B/16mean accuracy: 69.4 -> 70.7Improved
replace LoRA with AOFT* on ViT-LLoRA · VTAB-1k ViT-Lmean accuracy: 72.0 -> 74.3Improved
replace LoRA with AOFT* on ViT-HLoRA · VTAB-1k ViT-Hmean accuracy: 69.3 -> 71.3Improved
replace Adapter with AOFT* on ViT-LAdapter · VTAB-1k ViT-Lmean accuracy: 52.9 -> 60.9Improved
add learnable scaling vector (AOFT*) to LoRA+AOFT (ablation)LoRA+AOFT · VTAB-1k ViT-B/16mean accuracy: 72.9 -> 71.5Hurt
add learnable scaling vector (AOFT*) to Adapter+AOFT (ablation)Adapter+AOFT · FGVC ViT-B/16mean accuracy: 88.9 -> 90.1Improved
apply AOFT also to W_FC1 and W_FC2 in LoRA (ablation)LoRA+AOFT (Wq,Wv) · VTAB-1k ViT-B/16mean accuracy: 73.1 -> 73.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy