search

thuml / Deep-Embedded-Validation

Code release for Towards Accurate Model Selection in Deep Unsupervised Domain Adaptation (ICML 2019)
MIT License
61 stars 10 forks source link

Using DEV at other deep domain adaptation models #2

Closed userb2020 closed 4 years ago

userb2020 commented 5 years ago

Hello! I am really interested in your work, and I want to use your validation methods in my domain adaptation model. And I have several questions when I am using the function get_weight in dev.py

The three input parameters (source_feature, target_feature, validation_feature) should be calculated using whole dataset or per batch?

And if I calculate it per batch, what is the right size of the batch?

It would be very thankful if you answer my questions.

youkaichao commented 5 years ago

Thanks for your interest.

It should be calculated using the whole dataset.

If the dataset is too large, you should perhaps implement a mini-batch version yourself.