search

BCML-KW / Arrythmia-Classifier-using-GNN

GNN based Arrythmia classifier
0 stars 1 forks source link

dataset info #1

Closed shasha2408 closed 2 years ago

shasha2408 commented 2 years ago

Hello there can you please tell the location of the dataset ?

metr0jw commented 2 years ago

Hello there can you please tell the location of the dataset ?

You can put your own dataset into training/testing directory.

shasha2408 commented 2 years ago

Thank you so much for your response. I really appreciate that. Actually i am having trouble to implement gnn on multivariate time series anomaly detection. I found your repo at GitHub and tried to put my own dataset but there is some error in the output of the gnn. That’s why I asked if you can tell me which arrthymeya dataset you have used .

On Fri, Sep 2, 2022 at 9:16 PM Jiwoon Lee @.***> wrote:

Hello there can you please tell the location of the dataset ?

You can put your own dataset into training/testing directory.

— Reply to this email directly, view it on GitHub https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1235431070, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATRL6BKT5Z6VFHD6CKE67EDV4HV2RANCNFSM6AAAAAAQDEOOVI . You are receiving this because you authored the thread.Message ID: @.***>

-- Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

metr0jw commented 2 years ago

Thank you so much for your response. I really appreciate that. Actually i am having trouble to implement gnn on multivariate time series anomaly detection. I found your repo at GitHub and tried to put my own dataset but there is some error in the output of the gnn. That’s why I asked if you can tell me which arrthymeya dataset you have used .

On Fri, Sep 2, 2022 at 9:16 PM Jiwoon Lee @.***> wrote:

Hello there can you please tell the location of the dataset ?

You can put your own dataset into training/testing directory.

— Reply to this email directly, view it on GitHub https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1235431070, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATRL6BKT5Z6VFHD6CKE67EDV4HV2RANCNFSM6AAAAAAQDEOOVI . You are receiving this because you authored the thread.Message ID: @.***>

-- Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

Actually I'm not an author of this repository so I don't know much about it. But if you let me know what is the problem or the error message, I can forward it to the author.

shasha2408 commented 2 years ago

All right. I will let you know about the error tommorow. Again thank you so much for your response.

On Fri, Sep 2, 2022 at 9:30 PM Jiwoon Lee @.***> wrote:

Thank you so much for your response. I really appreciate that. Actually i am having trouble to implement gnn on multivariate time series anomaly detection. I found your repo at GitHub and tried to put my own dataset but there is some error in the output of the gnn. That’s why I asked if you can tell me which arrthymeya dataset you have used .

On Fri, Sep 2, 2022 at 9:16 PM Jiwoon Lee @.***> wrote:

Hello there can you please tell the location of the dataset ?

You can put your own dataset into training/testing directory.

— Reply to this email directly, view it on GitHub

1 (comment)

https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1235431070 , or unsubscribe

https://github.com/notifications/unsubscribe-auth/ATRL6BKT5Z6VFHD6CKE67EDV4HV2RANCNFSM6AAAAAAQDEOOVI . You are receiving this because you authored the thread.Message ID: @.***>

-- Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

Actually I'm not an author of this repository so I don't know much about it. But if you let me know what is the problem or the error message, I can forward it to the author.

— Reply to this email directly, view it on GitHub https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1235444305, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATRL6BLIVGUYKS6NXG6E3HTV4HXORANCNFSM6AAAAAAQDEOOVI . You are receiving this because you authored the thread.Message ID: @.***>

-- Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

shasha2408 commented 2 years ago

Good morning jeewon The issue I am facing here is that there are a lot of Arrhythmia datasets available. so i don't understand which dataset to put in training and testing. I also tried to put my dataset there but there is an error in graph making . so if i get to know the original dataset used in this repo. It will be very helpful for me. Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

On Fri, Sep 2, 2022 at 9:30 PM Jiwoon Lee @.***> wrote:

Thank you so much for your response. I really appreciate that. Actually i am having trouble to implement gnn on multivariate time series anomaly detection. I found your repo at GitHub and tried to put my own dataset but there is some error in the output of the gnn. That’s why I asked if you can tell me which arrthymeya dataset you have used .

On Fri, Sep 2, 2022 at 9:16 PM Jiwoon Lee @.***> wrote:

Hello there can you please tell the location of the dataset ?

You can put your own dataset into training/testing directory.

— Reply to this email directly, view it on GitHub

1 (comment)

https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1235431070 , or unsubscribe

https://github.com/notifications/unsubscribe-auth/ATRL6BKT5Z6VFHD6CKE67EDV4HV2RANCNFSM6AAAAAAQDEOOVI . You are receiving this because you authored the thread.Message ID: @.***>

-- Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

Actually I'm not an author of this repository so I don't know much about it. But if you let me know what is the problem or the error message, I can forward it to the author.

— Reply to this email directly, view it on GitHub https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1235444305, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATRL6BLIVGUYKS6NXG6E3HTV4HXORANCNFSM6AAAAAAQDEOOVI . You are receiving this because you authored the thread.Message ID: @.***>

kaintels commented 2 years ago

Thank you for reaching out to me. @metr0jw Good morning Shashank, @shasha2408

I extracted ECG feature (subject_000.npz) from MIT-BIH arrhythmia database.

After that, like the code below, I set the features and label to data["x"] and data["y"]. (There are many subjects in the path, so I used a for loop.)

I created a graph structure based on the correlation matrix from features. This research is being conducted experimentally, and I believe it should be improved.

If you have any questions, please send me an email or reply.

https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/blob/30cf103fcc9193577a501e29dab88b3edf3f7ef6/utils/util.py#L10-L13

Hello there can you please tell the location of the dataset ?

Best Regrads,

Seungwoo Han, AI Researcher, M.Sc

shasha2408 commented 2 years ago

Thank you again for your timely reply. Actually i tried to put my own dataset and preprocessed as you suggested but GNN output is zero after every epoch. I guess it is because of the architecture of the model and i/p, o/p layers. Is it possible if you can send me the Dataset and then i can try by myself and clear my doubts. it will be a great help and i will be highly thankful to you.

kaintels commented 2 years ago

Unfortunately, it is difficult to share the dataset because I graduated the lab.

Instead, I created a new repository with the breast cancer data set of sklearn. https://github.com/kaintels/GNN-classifier-test

https://github.com/kaintels/GNN-classifier-test/blob/9fcd08a897e1be424b47a1caf04c6aa758c6a649/utils/util.py#L16-L21

This GNN classifier classify using graph configured elements(include self-loop) of the Laplacian matrix(based correlation).

I used the correlation, but there are also graphs implemented using the dynamic warping method, take a look please.

paper name:

Thank you again for your timely reply. Actually i tried to put my own dataset and preprocessed as you suggested but GNN output is zero after every epoch. I guess it is because of the architecture of the model and i/p, o/p layers. Is it possible if you can send me the Dataset and then i can try by myself and clear my doubts. it will be a great help and i will be highly thankful to you.

Best Regrads,

Seungwoo Han, AI Researcher, M.Sc

shasha2408 commented 2 years ago

Thank you so much brother. Now I got it fully. I am really very very thankful to help me here bcoz i have been struggling for the last several days on this. kindly let me know if you need any help in future about everything. Thank you. Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea

On Mon, Sep 5, 2022 at 2:23 PM Seungwoo Han @.***> wrote:

Unfortunately, it is difficult to share the dataset because I graduated the lab.

Instead, I created a new repository with the breast cancer data set of sklearn. https://github.com/kaintels/GNN-classifier-test

  • Key Changes

https://github.com/kaintels/GNN-classifier-test/blob/9fcd08a897e1be424b47a1caf04c6aa758c6a649/utils/util.py#L16-L21

This GNN classifier classify using graph configured elements(include self-loop) of the Laplacian matrix.

I used the correlation, but there are also graphs implemented using the dynamic warping method, take a look please.

paper name:

  • Jiang, Li, et al. "Bearing fault diagnosis method based on a multi-head graph attention network." Measurement Science and Technology 33.7 (2022): 075012.

Thank you again for your timely reply. Actually i tried to put my own dataset and preprocessed as you suggested but GNN output is zero after every epoch. I guess it is because of the architecture of the model and i/p, o/p layers. Is it possible if you can send me the Dataset and then i can try by myself and clear my doubts. it will be a great help and i will be highly thankful to you.

Best Regrads,

Seungwoo Han, AI Researcher, M.Sc

— Reply to this email directly, view it on GitHub https://github.com/BCML-KW/Arrythmia-Classifier-using-GNN/issues/1#issuecomment-1236551194, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATRL6BNNUX4YADMQ4JGSDSTV4V7VDANCNFSM6AAAAAAQDEOOVI . You are receiving this because you were mentioned.Message ID: @.***>

kaintels commented 2 years ago

okay. I'm glad it helped. Have a good research.

Have a nice day,

Seungwoo Han, AI Researcher, M.Sc

Thank you so much brother. Now I got it fully. I am really very very thankful to help me here bcoz i have been struggling for the last several days on this. kindly let me know if you need any help in future about everything. Thank you. Thanks & Regards Shashank Chauhan MS Candidate, School of Industrial Engineering, University of Ulsan, Ulsan, South Korea