search

Fuminides / athena

Athena is a library that comprises many different bci frameworks that perform classification on a set of eeg data.
MIT License
19 stars 3 forks source link

Problem of running the example in this code #2

Open Moonshadowzb opened 3 years ago

Moonshadowzb commented 3 years ago

When I run the working example given in bci_architectures.py, I got a problem that the 'load_competition_subject' function needs data named'1TClass1_stft.mat' which I could not find.

I only have the BCI 2A source data named such as 'A01T.gdf' and label files. So how can I get the '1TClass1_stft.mat' data?

Fuminides commented 3 years ago

Hi!

The repository is (very) a preliminary work and somethings, like the data loading modules, are very precary. I hoped I could find time to keep improving the code, but recent events has made moved from this project further that I originally intended. (But I hope to find time again to work in this in a medium-range future)

The data files present in the code are the file formats that I peersonally used for my experiments. You can find a little bit of info about the expected data for each function in the documentation. I know that gdf is a very popular format, but I used .mat (MATLAB/Octave files) for this repository because it was easier for me. I think that you can export from the .gdf to .mat format. That would be the way to go.

Thank you.

Fuminides commented 3 years ago

Maybe it would be easier if you do the opposite path: adapt the bci architecture to the rest of your code. Since most of the architectures use fairly popular methods (CSP, LDA, etc) I think that you could reproduce them easily in other settings.

384068026 commented 2 years ago

When I run the working example given in bci_architectures.py, I got a problem that the 'load_competition_subject' function needs data named'1TClass1_stft.mat' which I could not find.

I only have the BCI 2A source data named such as 'A01T.gdf' and label files. So how can I get the '1TClass1_stft.mat' data?

Hello, I'm really sorry to disturb you. Have you solved the problem you raised in the comment section before? I'd like to ask you for some experience and hope to get your reply. My email is shichi1999js@163.com

384068026 commented 2 years ago

Hi!

The repository is (very) a preliminary work and somethings, like the data loading modules, are very precary. I hoped I could find time to keep improving the code, but recent events has made moved from this project further that I originally intended. (But I hope to find time again to work in this in a medium-range future)

The data files present in the code are the file formats that I peersonally used for my experiments. You can find a little bit of info about the expected data for each function in the documentation. I know that gdf is a very popular format, but I used .mat (MATLAB/Octave files) for this repository because it was easier for me. I think that you can export from the .gdf to .mat format. That would be the way to go.

Thank you.

Hi, I have been studying to reproduce your code recently, and I really want to get the data files like "A01tclass1_stft.mat" that you used at that time. Or can you tell me how to convert A01T.mat to A01TClass1_stft.mat?

Fuminides commented 2 years ago

This stft stantds for fast fourier transform. So what I called with the prefix *_stft there were the wavebands were a such transformation was applied.

zh0501 commented 2 years ago

I want to know what the dimension of the data after the fast Fourier transform looks like, can you tell me?For example (trials, number of channels, sample points)

384068026 commented 2 years ago

Hello, I have received your letter.

First of all,

I'm not the author of the project myself, I'm just a reader who used to debug the code. And because of the long time, some things are not very accurate memory, I do not know whether there is a problem with some of the things said, and I hope you can understand.

The shape of 2A data is AxBxC. The actual value is 1000x22x288.

A -- >1000, which is the data length of each event (4s *250Hz)

B -- >22, which represents the number of electrodes used, i.e. the number of channels

C -- >288, which represents the number of events, that is, the number of labels

------------------ 原始邮件 ------------------ 发件人: "Fuminides/athena" @.>; 发送时间: 2022年11月7日(星期一) 晚上9:00 @.>; @.**@.>; 主题: Re: [Fuminides/athena] Problem of running the example in this code (Issue #2)

I want to know what the dimension of the data after the fast Fourier transform looks like, can you tell me?For example (trials, number of channels, sample points)

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>