AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Beyond Static Allocation: Dynamic Sensitivity-Aware Fine-Tuning for Vision Transformers

Yuanyang Cao, Xichun Liu, Fuwei Zhang, Shangqi Deng, Ziyang Ren, Jianji Wang

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Efficient ViT adaptation

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
add dynamic reconfiguration over static allocation (ablation)DAF · VTAB-1kmean accuracy (%): 75.8 -> 76.4Improved
add context-aware sensitivity analysis (with LoRA in forward pass) instead of pristine backbone analysis (ablation)DAF · VTAB-1kmean accuracy (%): 74.6 -> 76.4Improved
add Rebuild-and-Refocus (freeze outdated, reallocate) instead of accumulating parameters (ablation)DAF · VTAB-1kmean accuracy (%): 75.3 -> 76.4Improved
use Taylor-based |g*w| sensitivity instead of random selection (ablation)DAF · VTAB-1kmean accuracy (%): 73.5 -> 76.4Improved
use |g*w| instead of gradient magnitude |g| (ablation)DAF · VTAB-1kmean accuracy (%): 75.0 -> 76.4Improved
reconfiguration interval of 10 epochs versus 5, 25 or 50 (ablation)DAF · VTAB-1kmean accuracy (%): 76.0 / 76.4 / 76.1 / 75.9 for 5/10/25/50 epochsMixed
increase parameter budget tau beyond 0.2 (ablation)DAF · VTAB-1kmean accuracy (%): 76.4 -> 76.2 (0.3), 76.1 (0.4)Hurt
decrease parameter budget tau from 0.2 to 0.1 (ablation)DAF · VTAB-1kmean accuracy (%): 76.4 -> 75.7Hurt
vary number of sensitivity batches M from 8 to 32 (ablation)DAF · VTAB-1kmean accuracy (%): within approximately 0.3%No clear effect
apply DAF to MAE and MoCo v3 self-supervised ViT backbonesDAF · VTAB-1k (MAE)mean accuracy (%): 69.2 -> 69.9 vs SPT-AdapterImproved
apply DAF to dense prediction (detection) versus LoRASwin-B Cascade Mask R-CNN · COCO val2017AP box: 50.4 -> 53.5Improved
apply DAF to semantic segmentation versus LoRASwin-L UperNet · ADE20KmIoU: 50.3 -> 52.0Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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