lecture_da

Lecture: Introduction to Domain Adaptation

This repository contains two scripts for educational purposes.

1. run_pytorch_basic.py: Standard deep-learning pipeline in Pytorch (used as an example for Empirical Risk Minimization)
2. run_shift_classification.py: Domain adaptation example using DANN algorithm and Importance Weighting (IW). The target dataset is the MNIST-M dataset first proposed in the DANN paper.

Slides are also included in the repo as a pdf file!

Run code

Make sure to get the necessary libraries like numpy and pytorch: pip install numpy matplotlib torch torchvision pot For importance weighting we use a few extra libraries: pip install scikit-learn quadprog scipy