AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

DeFT-LoRA: Decoupled and Fused Tuning with LoRA Experts for Universal Cross-Domain Retrieval

Ke Xu, Xiaozheng Shen, Shanshan Wang, Mengzhu Wang, Xun Yang

AAAI 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Feature calibration for unseen-class detection · 1 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 single LoRA with decoupled base+domain experts (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.5909 -> 0.6093Improved
replace naive LoRA-MoE with three-stage base+domain experts (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.5970 -> 0.6093Improved
add Base Expert (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.6014 -> 0.6093Improved
use dynamic gating instead of static averaging (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.6019 -> 0.6093Improved
share A matrix across experts instead of independent A (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.6070 -> 0.6093Improved
apply LoRA to deep layers 7-12 instead of shallow 1-6 (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.5558 -> 0.6068Improved
apply LoRA to all layers instead of only deep layers (ablation)DeFT-LoRA · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.6068 -> 0.6093Improved
add DeFT-LoRA to zero-shot CLIPCLIP ViT-B/32 · DomainNet UCDR (Infograph unseen, Unseen Gallery)mAP@200: 0.5007 -> 0.6093Improved
DeFT-LoRA vs UCDR-AdapterDeFT-LoRA · DomainNet UCDR (Painting unseen, Mixed Gallery)mAP@200: 0.7203 vs 0.7135Hurt
DeFT-LoRA vs ProS/UCDR-AdapterDeFT-LoRA · DomainNet UCDR (Sketch unseen, Mixed Gallery)mAP@200: 0.6073 vs 0.6014Hurt
DeFT-LoRA vs UCDR-AdapterDeFT-LoRA · DomainNet UCDR average (Unseen Gallery)mAP@200: 0.6071 -> 0.6212Improved
use LoRA experts instead of prompts (efficiency)DeFT-LoRA · DomainNet Infographtrainable parameters (M): 2.36 -> 1.29Improved

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