search

yongkyung-oh / Stable-Neural-SDEs

MIT License
15 stars 2 forks source link

Reproduction results #3

Open chenxiaodanhit opened 1 week ago

chenxiaodanhit commented 1 week ago

Thank you for your great contributions! I have a question regarding classification with irregular samples. I run the sepsis-sde.py with model name of Neural LSDE. However, the AUROC I achieved is only 82.92%, which is significantly lower than 90.0% reported in the paper. Could you please provide some details on how to accurately reprodece the results?

yongkyung-oh commented 1 week ago

Thank you for your comment. Do you have the same issue for the other models with optimal hyperparameters?

By the way, I'm working on refactoring the experiment code, especially sepsis data. I will let you know when it is ready.

chenxiaodanhit commented 6 days ago

Thank you for your patience! I encounter the same issue with the other models as well.