Why can't I reproduce the results reported in the paper?

[JiaxinYou97]

Hi, I’m very interested in the paper and the project you provided. I followed the code and method to detect sleep spindle on the database of MASS SS2 and tried to reproduce the reported results, making sure every detail is the same as instructed in the paper, but the result of f1 is only between 0.5-0.6 in test dataset, not 0.7-0.8 as reported.
I don’t know what’s wrong and why this happened. Hoping your reply, thanks very much!

[Mechanike]

楼上所以你最后复现出来了吗，我看你论文里面贴的dosed的结果还挺好的

[Chenh0332]

Hello, recently, I was replicating this paper. The data set training model provided by the author does not work well. And I want to use DREAMS or MASS data sets, how do you modify the corresponding format of the data set to match the data set format required for model training? Could you give me some suggestions? Thanks!

[Mechanike]

hi，do you have mass dataset？why i cant get the dataset while i submit the corresponding ethical files to Boreails

[Chenh0332]

hi，do you have mass dataset？why i cant get the dataset while i submit the corresponding ethical files to Boreails

I only have DREAMS Spindle and K-complexes dataset. Did your model training produce good results? After I modified the structure of the model, although the accuracy and f1-score were improved compared to the previous ones, they were still not high.

[Chenh0332]

hi，do you have mass dataset？why i cant get the dataset while i submit the corresponding ethical files to Boreails

I only have DREAMS Spindle and K-complexes dataset. Did your model training produce good results? After I modified the structure of the model, although the accuracy and f1-score were improved compared to the previous ones, they were still not high.

Do I need to modify the corresponding dataset format to meet the input requirements of the model for training with open datasets？

[Mechanike]

hi，do you have mass dataset？why i cant get the dataset while i submit the corresponding ethical files to Boreails

I only have DREAMS Spindle and K-complexes dataset. Did your model training produce good results? After I modified the structure of the model, although the accuracy and f1-score were improved compared to the previous ones, they were still not high.

dreams dataset is smaller than mass and i dont know whether it has an impact on the performance. and usually the model is trained on the mass dataset and tested on dreams and the performance on dreams is bad than mass. maybe its because of out of distribution problem or the labels is labeled by different people