-
Notifications
You must be signed in to change notification settings - Fork 153
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
About cross-domain-ccfnet #14
Comments
Did you use different embeddings for users in different domains? |
Oh, I know this assumption. In fact, I just follow your assumption. Maybe some problems happen in my code implement. What confused me is how to share user embedding across domain. My original idea is keep 'user_embeddings' variable unchanged.
To be honest, I just want to learn your model and try to mix with other knowledge to finish my Bachelor thesis, at the same time, I also want to lay the foundation for postgraduate study. So , I really appreciate you if you could help me fix these problems, it can save me much time to mix it with other knowledge. |
In your code above, I saw you assign different weights (parameters) in the neural network for user attribute embedding. |
Yes, that was a big mistakes. I have modifed my code(keep user_weight and user_bias unique) and the test RMSE can renduce regularly. But, the test RMSE is still higher than yours. When I run the ccf-net, the test RMSE roughly equal to train RMSE on the first iteration. While, In my code, the test RMSE is higher than train RMSE about 0.1(at least). Did I set the improper hyperparameters? Besides, Is it feasible what I have done(I mean the the whole idea)?
|
Hello, I have try to extend your ccfnet to cross-domain, but unfortunately, I find my model didn't work as good as ccfnet in single area. More specifically, the test RMSE is always high, I guess it may be suffering overfitting. As I have posted below, can you give me some instructions to help me better my code?
By the way, I just did what your paper said, spliting the Movielens movies into two disjointed set to simulate two area.
`def build_model(user_cf_feature_1, user_cf_feature_2, user_attr_feature_1, user_attr_feature_2, user_attr_rank,
item_cf_feature_1, item_cf_feature_2, item_attr_feature_1, item_attr_feature_2, item_attr_rank,
ratings_1, ratings_2, layer_size,
W_user, W_item_1, W_item_2,
W_user_attr, W_item_attr_1, W_item_attr_2,
lamb, lr, mu_1, mu_2):
The text was updated successfully, but these errors were encountered: