{
  "id": 77451,
  "title": "Dataset labelling",
  "url": "/competitions/vsb-power-line-fault-detection/discussion/77451",
  "author_name": "",
  "post_date": "2019-01-12T23:56:59.408606600Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hey all!\nFirst of all, thanks to the organizers for this great competition!\nIn topic \"Problem description\" Competition Host said that \"All of the signals were labeled by an human expert\". So in the competition the task is not  finding signals with PD pattern. The real task is finding  the signals that the human expert has labeled as signals with PD pattern. \nIf  the expert solved this task by visual analysis of the signals, may be machine learning is not needed for this task -- you may asks why the expert made the decision. Just program the expert logic!\nThe expert logic can be very hard, for example in image recognition we can't program easily the presence of  a cat in some photo. \nIn this problem, the logic of the expert is also very complex and we can not program this logic? or he just looking for some predefined set of PD patterns in the signal?</p>\n\n<p>Best regards, Vsevolod</p>",
  "messages": [
    {
      "id": "455086",
      "postDate": "01/12/2019 23:56:59",
      "content": "<p>Hey all!\nFirst of all, thanks to the organizers for this great competition!\nIn topic \"Problem description\" Competition Host said that \"All of the signals were labeled by an human expert\". So in the competition the task is not  finding signals with PD pattern. The real task is finding  the signals that the human expert has labeled as signals with PD pattern. \nIf  the expert solved this task by visual analysis of the signals, may be machine learning is not needed for this task -- you may asks why the expert made the decision. Just program the expert logic!\nThe expert logic can be very hard, for example in image recognition we can't program easily the presence of  a cat in some photo. \nIn this problem, the logic of the expert is also very complex and we can not program this logic? or he just looking for some predefined set of PD patterns in the signal?</p>\n\n<p>Best regards, Vsevolod</p>",
      "rawMarkdown": "Hey all!\nFirst of all, thanks to the organizers for this great competition!\nIn topic \"Problem description\" Competition Host said that \"All of the signals were labeled by an human expert\". So in the competition the task is not  finding signals with PD pattern. The real task is finding  the signals that the human expert has labeled as signals with PD pattern. \nIf  the expert solved this task by visual analysis of the signals, may be machine learning is not needed for this task -- you may asks why the expert made the decision. Just program the expert logic!\nThe expert logic can be very hard, for example in image recognition we can't program easily the presence of  a cat in some photo. \nIn this problem, the logic of the expert is also very complex and we can not program this logic? or he just looking for some predefined set of PD patterns in the signal?\n\nBest regards, Vsevolod",
      "votes": null
    },
    {
      "id": "455696",
      "postDate": "01/14/2019 11:34:29",
      "content": "<p>They wrote in some other thread that they checked after the expert, and it was PD in the lane, so it was also labelled by the \"de facto\" of having a discharge. However, more info how the humans label it would be helpful. Maybe organisers could suggest some useful links  -- relevant info.  So far looking at the data I have a problem to identify where the discharge might be. </p>",
      "rawMarkdown": "They wrote in some other thread that they checked after the expert, and it was PD in the lane, so it was also labelled by the \"de facto\" of having a discharge. However, more info how the humans label it would be helpful. Maybe organisers could suggest some useful links  -- relevant info.  So far looking at the data I have a problem to identify where the discharge might be.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 455696,
      "author_name": "blondinka",
      "author_url": "",
      "post_date": "01/14/2019 11:34:29",
      "content": "<p>They wrote in some other thread that they checked after the expert, and it was PD in the lane, so it was also labelled by the \"de facto\" of having a discharge. However, more info how the humans label it would be helpful. Maybe organisers could suggest some useful links  -- relevant info.  So far looking at the data I have a problem to identify where the discharge might be. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "455086": "Hey all!\nFirst of all, thanks to the organizers for this great competition!\nIn topic \"Problem description\" Competition Host said that \"All of the signals were labeled by an human expert\". So in the competition the task is not  finding signals with PD pattern. The real task is finding  the signals that the human expert has labeled as signals with PD pattern. \nIf  the expert solved this task by visual analysis of the signals, may be machine learning is not needed for this task -- you may asks why the expert made the decision. Just program the expert logic!\nThe expert logic can be very hard, for example in image recognition we can't program easily the presence of  a cat in some photo. \nIn this problem, the logic of the expert is also very complex and we can not program this logic? or he just looking for some predefined set of PD patterns in the signal?\n\nBest regards, Vsevolod",
    "455696": "They wrote in some other thread that they checked after the expert, and it was PD in the lane, so it was also labelled by the \"de facto\" of having a discharge. However, more info how the humans label it would be helpful. Maybe organisers could suggest some useful links  -- relevant info.  So far looking at the data I have a problem to identify where the discharge might be."
  },
  "source": "meta"
}