Reducing the Covariate Shift by Mirror Samples in Cross Domain Alignment

Authors: Yin Zhao, minquan wang, Longjun Cai

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

Reproducibility Variable Result LLM Response
Research Type Experimental By applying the virtual mirror and mirror loss to the generic unsupervised domain adaptation model, we achieved consistently superior performance on several mainstream benchmarks.
Researcher Affiliation Industry Yin Zhao Alibaba Group yinzhao.zy@alibaba-inc.com Minquan Wang Alibaba Group minquan.wmq@alibaba-inc.com Longjun Cai Alibaba Group longjun.clj@alibaba-inc.com
Pseudocode Yes The detailed algorithm can be found in Appendix.
Open Source Code No The paper does not provide any statement or link indicating that its source code is open or publicly available.
Open Datasets Yes Datasets. We use Office-31 [42], Office-Home[51], Image CLEF and Vis DA2017[39] to validate our proposed method.
Dataset Splits No The paper mentions using 'validation' as a target domain for VisDA2017, but it does not provide explicit percentages or sample counts for training, validation, or test splits for any of the datasets used in the main text.
Hardware Specification Yes All the experiments are carried out on one Tesla V100 GPU.
Software Dependencies No We implement our model in Py Torch. (No version specified for PyTorch or other libraries)
Experiment Setup Yes The learning rate is adjusted by ηp = η0(1 + αp) β like [17], where p is the epoch which is normalized in [0, 1], η0 = 0.001, α = 10 and β = 0.75. The learning rate of fully connected layers is 10 times of the backbone layers. [...] (K = 3) [...] λ = 1.0.