search

MousaviSajad / SleepEEGNet

SleepEEGNet: Automated Sleep Stage Scoring with Sequence to Sequence Deep Learning Approach
200 stars 72 forks source link

Hi. I have a question. #9

Open hji1014 opened 4 years ago

hji1014 commented 4 years ago

Hello. I'm a graduate student in Korea. I would like to get the same result of accuracy:84.26%, kappa coefficient:0.79. However, changing the number of epochs, changing the batch_size, changing the optimizer type, turning on or turning off bidirectional produces low results. Please give me some advice. Thank you!

MousaviSajad commented 4 years ago

Hi, To get the same result, you should use the same values for the hyperparameters that I have used.

Regards, Sajad

On Mon, May 25, 2020 at 9:27 PM JUNILLHUH notifications@github.com wrote:

Hello. I'm a graduate student in Korea. I would like to get the same result of accuracy:84.26%, kappa coefficient:0.79. However, changing the number of epochs, changing the batch_size, changing the optimizer type, turning on or turning off bidirectional produces low results. Please give me some advice. Thank you!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/MousaviSajad/SleepEEGNet/issues/9, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD4D7SM3HT55MZ2RJOI6PDLRTMLI3ANCNFSM4NJ4LQYQ .

hji1014 commented 4 years ago

Thank you for your answer.

When I ran the code with the parameters in your paper, the overall accuracy was about 5% lower.

Do I need to change 'LOSS FUNCTION' to MFE or MSFE in your paper?

I'm sorry for asking you a low-level question because I just started studying.

Thank you MousaviSajad.

-----Original Message----- From: "Sajad Mousavi"notifications@github.com To: "MousaviSajad/SleepEEGNet"SleepEEGNet@noreply.github.com; Cc: "JUNILLHUH"hji10141@naver.com; "Author"author@noreply.github.com; Sent: 2020-05-26 (화) 23:22:09 (GMT+09:00) Subject: Re: [MousaviSajad/SleepEEGNet] Hi. I have a question. (#9)

Hi, To get the same result, you should use the same values for the hyperparameters that I have used.

Regards, Sajad

On Mon, May 25, 2020 at 9:27 PM JUNILLHUH notifications@github.com wrote:

Hello. I'm a graduate student in Korea. I would like to get the same result of accuracy:84.26%, kappa coefficient:0.79. However, changing the number of epochs, changing the batch_size, changing the optimizer type, turning on or turning off bidirectional produces low results. Please give me some advice. Thank you!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/MousaviSajad/SleepEEGNet/issues/9, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD4D7SM3HT55MZ2RJOI6PDLRTMLI3ANCNFSM4NJ4LQYQ .

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

hji1014 commented 4 years ago

Your code says 'bidirectional' hyperparameter--> 'False', do you mean to run it with 'False'?

Thank you.

capture

MousaviSajad commented 4 years ago

Hi, 'bidirectional' hyperparameter--> 'False' means the code is not going to use the bidirectional RNN. Regards, Sajad

On Tue, Jun 2, 2020 at 4:39 AM JUNILLHUH notifications@github.com wrote:

Your code says 'bidirectional' hyperparameter--> 'False', do you mean to run it with 'False'?

Thank you.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/MousaviSajad/SleepEEGNet/issues/9#issuecomment-637386517, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD4D7SK26GKEIZC6ILXBFZDRUS3FZANCNFSM4NJ4LQYQ .

shenjianaixuexi commented 4 years ago

Dear Sajad： I have the same questions .I do the experiment with your code and hyperparameter many times ,but I can't get the same results as you provided.What should i do next ? Regards, Jianji

nengwp commented 4 years ago

您收到此消息是因为您创建了线程。直接回复此电子邮件，在GitHub上查看或取消订

I also used the default parameters provided by the author, but the performance was also reduced by about 5%. Have you reached the results in the paper? thank you very much

nengwp commented 4 years ago

嗨，为了获得相同的结果，您应该为我使用的超参数使用相同的值。问候，萨哈德 … 于2020年5月25日星期一9:27 PM JUNILLHUH @.***>写道：您好。我是韩国的研究生。我想得到相同结果的准确性：84.26％，卡伯系数：0.79。但是，更改时期数，更改batch_size，更改优化器类型，打开或关闭双向都会产生较低的结果。请给我一些建议。谢谢！—您收到此消息是因为您已订阅此线程。直接回复此电子邮件，在GitHub < ＃9 > 上查看，或取消订阅< https://github.com/notifications/unsubscribe-auth/AD4D7SM3HT55MZ2RJOI6PDLRTMLI3ANCNFSM4NJ4LQYQ >。

I don't know if you use your parameters, can you give your default parameters, because your parameters in the code cannot achieve the effect of the paper.

LindseyScharpman commented 3 years ago

Dear Sajad： I have the same questions .I do the experiment with your code and hyperparameter many times ,but I can't get the same results as you provided.What should i do next ? Regards, Jianji

HI, I have the same problem, Do u solve it ?

nengwp commented 3 years ago

Dear Sajad： I have the same questions .I do the experiment with your code and hyperparameter many times ,but I can't get the same results as you provided.What should i do next ? Regards, Jianji

HI, I have the same problem, Do u solve it ?

This is probably a piece of paper that can't be repeated, but we published a paper that went beyond that