{ "id": 362816, "title": "Is there an available dataset for me to test the training? (and it contains official data)", "url": "/competitions/g2net-detecting-continuous-gravitational-waves/discussion/362816", "author_name": "StarxSky", "post_date": "2022-10-29T08:36:25.366000", "votes": 9, "comment_count": 2, "views": 0, "content": "

Hello \"G2Neter\" ！！ πŸ˜„Here is a dataset generated by my own adjustment of parameters using pyfstat It contains the path to the official dataset file and you can use it!βœ”οΈ (This dataset does not provide the source code for the generated data)

\n

However, you can use it to test the performance of your model and good luck with your results!!! Also hope this dataset will be useful for your work! Please vote for this dataset if you think it is good!πŸ™πŸ™

\n
\n

Dataset

\n
", "messages": [ { "id": 2008680, "postDate": "2022-10-29T08:36:25.367Z", "content": "

Hello \"G2Neter\" ！！ πŸ˜„Here is a dataset generated by my own adjustment of parameters using pyfstat It contains the path to the official dataset file and you can use it!βœ”οΈ (This dataset does not provide the source code for the generated data)

\n

However, you can use it to test the performance of your model and good luck with your results!!! Also hope this dataset will be useful for your work! Please vote for this dataset if you think it is good!πŸ™πŸ™

\n
\n

Dataset

\n
", "rawMarkdown": "#### Hello \"G2Neter\" ！！ πŸ˜„Here is a dataset generated by my own adjustment of parameters using pyfstat It contains the path to the official dataset file and you can use it!βœ”οΈ (This dataset does not provide the source code for the generated data)\n#### However, you can use it to test the performance of your model and good luck with your results!!! Also hope this dataset will be useful for your work! Please vote for this dataset if you think it is good!πŸ™πŸ™\n\n> **[Dataset](https://www.kaggle.com/datasets/lau01b/g2netsignalgenerate)**\n\n", "votes": 9 }, { "id": 2022141, "postDate": "2022-11-08T18:26:26.270Z", "content": "

Hi @lau01b, I tested your dataset in my model and found no accuracy improvement compared to the original train dataset. This makes me think that besides generating more data like the training dataset, this competition requires other data augmentation tricks. E.g. matching the deviation from gaussian noise like in the test dataset or accounting for glitches and off phases of the detectors. What kind of data augmentation techniques did you use to achieve your LB score?

", "rawMarkdown": "Hi @lau01b, I tested your dataset in my model and found no accuracy improvement compared to the original train dataset. This makes me think that besides generating more data like the training dataset, this competition requires other data augmentation tricks. E.g. matching the deviation from gaussian noise like in the test dataset or accounting for glitches and off phases of the detectors. What kind of data augmentation techniques did you use to achieve your LB score?", "votes": 3 }, { "id": 2023142, "postDate": "2022-11-09T15:04:23.350Z", "content": "

@simonebvr Hello! I'm glad to receive your letter. Firstly, you can try the following mixup data enhancement methods. Secondly, I need to adjust this data set later, because its data distribution may not be correct.πŸ™πŸ™

", "rawMarkdown": "@simonebvr Hello! I'm glad to receive your letter. Firstly, you can try the following mixup data enhancement methods. Secondly, I need to adjust this data set later, because its data distribution may not be correct.πŸ™πŸ™" } ], "comments": [ { "id": 2022141, "author_name": "Simone Bavera", "author_url": "", "post_date": "2022-11-08T18:26:26.270000", "content": "

Hi @lau01b, I tested your dataset in my model and found no accuracy improvement compared to the original train dataset. This makes me think that besides generating more data like the training dataset, this competition requires other data augmentation tricks. E.g. matching the deviation from gaussian noise like in the test dataset or accounting for glitches and off phases of the detectors. What kind of data augmentation techniques did you use to achieve your LB score?

", "votes": 3, "replies": [] }, { "id": 2023142, "author_name": "StarxSky", "author_url": "", "post_date": "2022-11-09T15:04:23.350000", "content": "

@simonebvr Hello! I'm glad to receive your letter. Firstly, you can try the following mixup data enhancement methods. Secondly, I need to adjust this data set later, because its data distribution may not be correct.πŸ™πŸ™

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2008680": "#### Hello \"G2Neter\" ！！ πŸ˜„Here is a dataset generated by my own adjustment of parameters using pyfstat It contains the path to the official dataset file and you can use it!βœ”οΈ (This dataset does not provide the source code for the generated data)\n#### However, you can use it to test the performance of your model and good luck with your results!!! Also hope this dataset will be useful for your work! Please vote for this dataset if you think it is good!πŸ™πŸ™\n\n> **[Dataset](https://www.kaggle.com/datasets/lau01b/g2netsignalgenerate)**\n\n", "2022141": "Hi @lau01b, I tested your dataset in my model and found no accuracy improvement compared to the original train dataset. This makes me think that besides generating more data like the training dataset, this competition requires other data augmentation tricks. E.g. matching the deviation from gaussian noise like in the test dataset or accounting for glitches and off phases of the detectors. What kind of data augmentation techniques did you use to achieve your LB score?", "2023142": "@simonebvr Hello! I'm glad to receive your letter. Firstly, you can try the following mixup data enhancement methods. Secondly, I need to adjust this data set later, because its data distribution may not be correct.πŸ™πŸ™" } }