Is there a way to handle EEG signals? I tried padding the signals so it can load as batches, but GPUs run out of memory when loading them. So is there a way to preprocess to make them smaller with minimum loss of information?
", "messages": [ { "id": "2738909", "postDate": "04/06/2024 17:38:17", "content": "Is there a way to handle EEG signals? I tried padding the signals so it can load as batches, but GPUs run out of memory when loading them. So is there a way to preprocess to make them smaller with minimum loss of information?
", "rawMarkdown": "Is there a way to handle EEG signals? I tried padding the signals so it can load as batches, but GPUs run out of memory when loading them. So is there a way to preprocess to make them smaller with minimum loss of information?", "votes": null }, { "id": "2738958", "postDate": "04/06/2024 18:11:15", "content": "All subsamples are equally long (50 seconds - 10_000 time steps), no need of paddings. A single parquet contains various different subsamples. Check https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468010
", "rawMarkdown": "All subsamples are equally long (50 seconds - 10_000 time steps), no need of paddings. A single parquet contains various different subsamples. Check https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468010", "votes": null }, { "id": "2739007", "postDate": "04/06/2024 19:07:08", "content": "I wish I knew about this before… feeling guilty that I wasted my time on it… thank you for the help!
", "rawMarkdown": "I wish I knew about this before... feeling guilty that I wasted my time on it... thank you for the help!", "votes": null }, { "id": "2739693", "postDate": "04/07/2024 07:55:01", "content": "There is this dataset from Chris and used in his starter notebook which should also have code for creating one if you want more features.
", "rawMarkdown": "There is [this dataset](https://www.kaggle.com/datasets/cdeotte/brain-eegs) from Chris and used in his [starter notebook](https://www.kaggle.com/code/cdeotte/wavenet-starter-lb-0-66) which should also have code for creating one if you want more features.", "votes": null } ], "comments": [ { "id": 2738958, "author_name": "sacuscreed", "author_url": "", "post_date": "04/06/2024 18:11:15", "content": "All subsamples are equally long (50 seconds - 10_000 time steps), no need of paddings. A single parquet contains various different subsamples. Check https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468010
", "votes": null, "replies": [ { "id": 2739007, "author_name": "anrenk", "author_url": "", "post_date": "04/06/2024 19:07:08", "content": "I wish I knew about this before… feeling guilty that I wasted my time on it… thank you for the help!
", "votes": null, "replies": [] } ] }, { "id": 2739693, "author_name": "something4kag", "author_url": "", "post_date": "04/07/2024 07:55:01", "content": "There is this dataset from Chris and used in his starter notebook which should also have code for creating one if you want more features.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2738909": "Is there a way to handle EEG signals? I tried padding the signals so it can load as batches, but GPUs run out of memory when loading them. So is there a way to preprocess to make them smaller with minimum loss of information?", "2738958": "All subsamples are equally long (50 seconds - 10_000 time steps), no need of paddings. A single parquet contains various different subsamples. Check https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468010", "2739007": "I wish I knew about this before... feeling guilty that I wasted my time on it... thank you for the help!", "2739693": "There is [this dataset](https://www.kaggle.com/datasets/cdeotte/brain-eegs) from Chris and used in his [starter notebook](https://www.kaggle.com/code/cdeotte/wavenet-starter-lb-0-66) which should also have code for creating one if you want more features." }, "source": "meta" }