Some questions regarding the source code

[fbiying87]

To the author of the paper "Stratified Transfer Learning for Cross-domain Activity Recognition.":

I come across your work in cross domain adaptation for activity recognition and find it very interesting to read. I was just wondering, if you have the source code also available in python or pytorch?

Other questions regarding the paper:
1.) Does the source and target domain have the number of samples for each classes?
2.) Should the source and target domain have the same amount of samples?
3.) Do you update the mmd_loss for each batch or the entire dataset?
4.) Can you maybe also share the data with me? I can't find the matlab matrix for dsads.mat.

Thanks for your reply!

Best regards,
Biying

[jindongwang]

Hi @fbiying87, sorry I just see this issue now since this repo is not very active recently.
As for the code for STL, I only have the matlab version and there is no python version. This paper is a very old work of mine (done in 2016), therefore a lot of details are missing especially the code. And since I have also proposed other transfer learning methods which are better than STL, I strongly suggest you visit transferlearning.xyz to see other algorithms.
Regarding your questions:

1. Yes. Currently it is a homogeneous transfer learning problem, which of course requires that both domains have the same number of classes.
2. No. Both domains do not need to have the same number of samples.
3. This is not a deep method, therefore there is no such concept about 'batch' or update. It is done in a single iteration. Further iterations are only to ensure better results.
4. The data are lost too since I have graduated, and the machine is returned to the PhD lab. But I have published the data to Kaggle: https://www.kaggle.com/jindongwang92/crossposition-activity-recognition