I have attached the full list of names assigned to wsi so there are no more unknowns.
", "messages": [ { "id": 496876, "postDate": "2019-03-22T17:39:11.177Z", "content": "I have attached the full list of names assigned to wsi so there are no more unknowns.
", "rawMarkdown": " I have attached the full list of names assigned to wsi so there are no more unknowns.", "votes": 12 }, { "id": 499241, "postDate": "2019-03-24T14:32:59.313Z", "content": "Hi All, I shared this in good faith to enable focus to move on from WSI assignment. I apologize for any part this had in encouraging cheats to submit fake scores and ruin the competition. I don't really understand what people gain from cheating their way to the top.
", "rawMarkdown": "Hi All, I shared this in good faith to enable focus to move on from WSI assignment. I apologize for any part this had in encouraging cheats to submit fake scores and ruin the competition. I don't really understand what people gain from cheating their way to the top.", "votes": 4 }, { "id": 497197, "postDate": "2019-03-23T05:11:52.237Z", "content": "Nice job. Did it give you any improvement?
", "rawMarkdown": "Nice job. Did it give you any improvement? ", "votes": 1, "replies": [ { "id": 497350, "postDate": "2019-03-23T12:15:35.553Z", "content": "Hi, Ivan. This WSI files really works. My score of single model went up from 0.9720 to 0.9752 using this file.
", "rawMarkdown": "Hi, Ivan. This WSI files really works. My score of single model went up from 0.9720 to 0.9752 using this file.", "votes": 3 }, { "id": 497646, "postDate": "2019-03-23T18:58:19.713Z", "content": "Does \"...normal...\" mean that there is no cancer? Thank you!
", "rawMarkdown": "Does \"...normal...\" mean that there is no cancer? Thank you!" } ] }, { "id": 497665, "postDate": "2019-03-23T19:14:15.600Z", "content": "Hello!\nIf I compare the \"old\" train labels with your \"wsi\" labels, so there are \n- 89117 labels with cancer and\n- 130908 labels with no cancer in the \"old\" train set.\nAnd in your \"wsi\" set, there are\n- 137729 labels with cancer and\n- 82296 labels with no cancer assuming that \"normal\" means no cancer and \"tumor\" means cancer.\nDoes it make sense?\nThank you!
", "rawMarkdown": "Hello!\nIf I compare the \"old\" train labels with your \"wsi\" labels, so there are \n- 89117 labels with cancer and\n- 130908 labels with no cancer in the \"old\" train set.\nAnd in your \"wsi\" set, there are\n- 137729 labels with cancer and\n- 82296 labels with no cancer assuming that \"normal\" means no cancer and \"tumor\" means cancer.\nDoes it make sense?\nThank you!", "replies": [ { "id": 497714, "postDate": "2019-03-23T20:30:54.867Z", "content": "the word tumor and normal are the labels for the original WSI, not the patches from this competition's dataset.
Hello Hanke, thank you!\nDon't understand. How should be then used this list derived from wsi?
", "rawMarkdown": "Hello Hanke, thank you!\nDon't understand. How should be then used this list derived from wsi? " }, { "id": 497773, "postDate": "2019-03-23T22:18:02.743Z", "content": "Yep, don’t change the patch labels to the wsi label.
", "rawMarkdown": "Yep, don’t change the patch labels to the wsi label. " }, { "id": 497824, "postDate": "2019-03-24T01:02:37.023Z", "content": "To create a stable cv split. But since the competition is over: (see: https://www.kaggle.com/c/histopathologic-cancer-detection/discussion/85424) you don't have to worry about that now
", "rawMarkdown": "To create a stable cv split. But since the competition is over: (see: https://www.kaggle.com/c/histopathologic-cancer-detection/discussion/85424) you don't have to worry about that now" }, { "id": 497830, "postDate": "2019-03-24T01:20:59.167Z", "content": "It is just a playground, but I'll operate under the assumption that the hashing trick submissions will be disregarded. Anyway, not bad finding it, I was waiting for someone to post it.
", "rawMarkdown": "It is just a playground, but I'll operate under the assumption that the hashing trick submissions will be disregarded. Anyway, not bad finding it, I was waiting for someone to post it." } ] }, { "id": 497617, "postDate": "2019-03-23T18:17:48.950Z", "rawMarkdown": "", "isDeleted": true }, { "id": 496955, "postDate": "2019-03-22T19:27:04.183Z", "content": "Thank you
", "rawMarkdown": "Thank you\n" }, { "id": 496938, "postDate": "2019-03-22T18:52:31.437Z", "content": "Thanks for sharing!
", "rawMarkdown": "Thanks for sharing!" } ], "comments": [ { "id": 499241, "author_name": "Mark De Simone", "author_url": "", "post_date": "2019-03-24T14:32:59.313000", "content": "Hi All, I shared this in good faith to enable focus to move on from WSI assignment. I apologize for any part this had in encouraging cheats to submit fake scores and ruin the competition. I don't really understand what people gain from cheating their way to the top.
", "votes": 4, "replies": [] }, { "id": 497197, "author_name": "Ivan Panshin", "author_url": "", "post_date": "2019-03-23T05:11:52.237000", "content": "Nice job. Did it give you any improvement?
", "votes": 1, "replies": [ { "id": 497350, "author_name": "seefun", "author_url": "", "post_date": "2019-03-23T12:15:35.553000", "content": "Hi, Ivan. This WSI files really works. My score of single model went up from 0.9720 to 0.9752 using this file.
", "votes": 3, "replies": [] }, { "id": 497646, "author_name": "neuralartist", "author_url": "", "post_date": "2019-03-23T18:58:19.713000", "content": "Does \"...normal...\" mean that there is no cancer? Thank you!
", "votes": 0, "replies": [] } ] }, { "id": 497665, "author_name": "neuralartist", "author_url": "", "post_date": "2019-03-23T19:14:15.600000", "content": "Hello!\nIf I compare the \"old\" train labels with your \"wsi\" labels, so there are \n- 89117 labels with cancer and\n- 130908 labels with no cancer in the \"old\" train set.\nAnd in your \"wsi\" set, there are\n- 137729 labels with cancer and\n- 82296 labels with no cancer assuming that \"normal\" means no cancer and \"tumor\" means cancer.\nDoes it make sense?\nThank you!
", "votes": 0, "replies": [ { "id": 497714, "author_name": "Hanke Chen", "author_url": "", "post_date": "2019-03-23T20:30:54.867000", "content": "the word tumor and normal are the labels for the original WSI, not the patches from this competition's dataset.
Hello Hanke, thank you!\nDon't understand. How should be then used this list derived from wsi?
", "votes": 0, "replies": [] }, { "id": 497773, "author_name": "interneuron", "author_url": "", "post_date": "2019-03-23T22:18:02.743000", "content": "Yep, don’t change the patch labels to the wsi label.
", "votes": 0, "replies": [] }, { "id": 497824, "author_name": "Hanke Chen", "author_url": "", "post_date": "2019-03-24T01:02:37.023000", "content": "To create a stable cv split. But since the competition is over: (see: https://www.kaggle.com/c/histopathologic-cancer-detection/discussion/85424) you don't have to worry about that now
", "votes": 0, "replies": [] }, { "id": 497830, "author_name": "interneuron", "author_url": "", "post_date": "2019-03-24T01:20:59.167000", "content": "It is just a playground, but I'll operate under the assumption that the hashing trick submissions will be disregarded. Anyway, not bad finding it, I was waiting for someone to post it.
", "votes": 0, "replies": [] } ] }, { "id": 497617, "author_name": "", "author_url": "", "post_date": "2019-03-23T18:17:48.950000", "content": "", "votes": 0, "replies": [] }, { "id": 496955, "author_name": "William Green", "author_url": "", "post_date": "2019-03-22T19:27:04.183000", "content": "Thank you
", "votes": 0, "replies": [] }, { "id": 496938, "author_name": "Joni Juvonen", "author_url": "", "post_date": "2019-03-22T18:52:31.437000", "content": "Thanks for sharing!
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "496876": " I have attached the full list of names assigned to wsi so there are no more unknowns.", "499241": "Hi All, I shared this in good faith to enable focus to move on from WSI assignment. I apologize for any part this had in encouraging cheats to submit fake scores and ruin the competition. I don't really understand what people gain from cheating their way to the top.", "497197": "Nice job. Did it give you any improvement? ", "497665": "Hello!\nIf I compare the \"old\" train labels with your \"wsi\" labels, so there are \n- 89117 labels with cancer and\n- 130908 labels with no cancer in the \"old\" train set.\nAnd in your \"wsi\" set, there are\n- 137729 labels with cancer and\n- 82296 labels with no cancer assuming that \"normal\" means no cancer and \"tumor\" means cancer.\nDoes it make sense?\nThank you!", "497617": "", "496955": "Thank you\n", "496938": "Thanks for sharing!" } }