AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Efficient Transfer Learning via Joint Adaptation of Network Architecture and Weight

Ming Sun, Haoxuan Dou, Junjie Yan

ECCV 2020 · Architectures and efficiency › Neural architecture search · approach: One-shot and few-shot supernets · 5 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
search the block allocation of ResNet50 on the target task via transfer (to [1,3,7,5])Faster-RCNN FPN ResNet50 · MS-COCO minivalmAP: 36.4 -> 37.8Improved
search the block allocation of ResNet101 on the target taskFaster-RCNN FPN ResNet101 · MS-COCO minivalmAP: 38.5 -> 39.8Improved
search MobileNetV2 block allocation on the target taskMobileNetV2 · MS-COCO minivalmAP: 32.2 -> 34.0Improved
search MnasNet-b0 block allocation on the target taskMnasNet-b0 · MS-COCO minivalmAP: 34.2 -> 34.7Improved
use full-space greedy search with a single longest-path super-network instead of the reduced search space of prior work (CR-ResNet50)ResNet50 · MS-COCO minivalmAP: 37.4 -> 37.8Improved
use transferred ResNet50 backbone within Grid-RCNNGrid-RCNN-ResNet50 · MS-COCO minivalmAP: 39.5 -> 40.3Improved
use transferred ResNet101 backbone within Cascade-MaskRCNNCascade-MaskRCNN-ResNet101 · MS-COCO minivalmAP: 43.3 -> 44.1Improved
transfer architecture to SE-ResNeXt50 with multi-branching and SE modulesSE-ResNeXt50 · MS-COCO minivalmAP: 38.9 -> 39.9Improved
increase computation budget of the transferred ResNet from ResNet37 to ResNet53 (ablation)Transfer-ResNet · MS-COCO minivalmAP: 36.4 -> 38.0Improved
transfer ResNet50 architecture to [1,6,6,3] on CUB-200-2011ResNet50 · CUB-200-2011mAP: 84.05 -> 84.98Improved
transfer architecture with non-local module on CUB-200-2011ResNet50+NL · CUB-200-2011mAP: 84.79 -> 85.42Improved
transfer ResNet50 architecture to [1,6,6,3] on CUB-200-2010ResNet50 · CUB-200-2010mAP: 68.01 -> 71.35Improved

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