Combating Noise: Semi-supervised Learning by Region Uncertainty Quantification

Authors: Zhenyu Wang, Ya-Li Li, Ye Guo, Shengjin Wang

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on both PASCAL VOC and MS COCO demonstrate the extraordinary performance of our method. Our method achieves the state-of-the-art results on the PASCAL VOC and MS COCO dataset, exceeding supervised baseline methods by 6.2% and 4.2% respectively.
Researcher Affiliation Academia Zhenyu Wang Yali Li Ye Guo Shengjin Wang Beijing National Research Center for Information Science and Technology (BNRist) Department of Electronic Engineering, Tsinghua University {wangzy20, guo-y18}@mails.tsinghua.edu.cn {liyali13, wgsgj}@tsinghua.edu.cn
Pseudocode No The paper describes the method through text and equations but does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement about open-sourcing its code or a direct link to a code repository for its described methodology.
Open Datasets Yes Dataset. We mainly conduct our proposed method on PASCAL VOC [6] and MS COCO [22].
Dataset Splits Yes We mainly adopt four settings: 1) VOC07 trainval (5,011 images) as labeled set and VOC12 trainval (11,540 images) as unlabeled set; 2) VOC07 trainval as labeled set, VOC12 trainval and coco-20cls as unlabeled set; 3) coco-35 as labeled set and coco-80 as labeled set; 4) coco-115 as labeled set and coco-120 as unlabeled set.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or specific computing environments with specs) used for running its experiments.
Software Dependencies No The paper mentions 'Pytorch [29] and MMDetection [5]' but does not specify their version numbers, which are necessary for reproducibility.
Experiment Setup Yes For hyper-parameters, we set b to 0.5. f is set to 0.7 first, and changes to 0.8 after the first decay of learning rate. C is set to 15 and q is set to 0.1.