goiter / CoCPC

49 stars 17 forks source link

Use of multi-task learning #1

Open shisi-cc opened 2 years ago

shisi-cc commented 2 years ago

I am a little confused about the objective function. In the paper you said that the total objective function is to minimize formula (15) (Kendall, Gal, and Cipolla 2018), but the code seems just simply adding up the two losses. Am I wrong or are there other considerations?

goiter commented 2 years ago

Thank you for commenting on this issue. In this code, we define the weights of each loss as 1 by testing on varied hyperparameter values. If you are interested in the adaptive weighting method, please try it.