STEM: An approach to Multi-source Domain Adaptation with Guarantees

Introduction

This is the official implementation of “STEM: An approach to Multi-source Domain Adaptation with Guarantees”

Prerequisites

System Requirement:

• Anaconda3
• Cuda toolkit 10.0

Install other environment requirement by Anaconda3 following:

conda env create -f env.yml

Note: the environment requires tensorbayes libs, however, available tensorbayes using Python 2.7. To fix the problem, please download tensorbayes, untar it and override to the env (stem) folder:

tar -xvf tensorbayes.tar
cp -rf ./tensorbayes/* /opt/conda/envs/stem/lib/python3.6/site-packages/tensorbayes/

Dataset Preparation

Please download and unzip the dataset and save all *.mat file under ../datasets. To save time computing, we extracted ResNet101 feature for Office-Caltech10 and provided as following:

• Digits-five
• Office-31
• DomainNet

Training

The config parameter to train model in config folder, please check it before run. To train the model, simply run:

<div class="snippet-clipboard-content position-relative overflow-auto" data-snippet-clipboard-copy-content="python run_stem_ht_mimic_hs.py –config .yaml –trg_name “>

python run_stem_ht_mimic_hs.py --config 
   
    .yaml --trg_name