{
  "id": 264561,
  "title": "Can we learn something from SETI?",
  "url": "/competitions/g2net-gravitational-wave-detection/discussion/264561",
  "author_name": "",
  "post_date": "2021-08-12T13:03:27.850047600Z",
  "votes": 3,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I was playing with the data while I noticed that if we set:<br>\n<code>Spectrogram.CQT1992v2(sr=2048, fmin=20, fmax=1024, hop_length=5, bins_per_octave=48, pad_mode='constant')</code><br>\nwe can get a spectrogram that has  <code>freq_bins=273</code> amd <code>time_steps=820</code> (819 + 1 for appox) that is the <strong>same spectrum as the SETI challenge</strong>.</p>\n<p>There is <strong>one big downside</strong> CQT is quite expensive at this resolution. 3h/epoch regardless of GPU/CPU with my hardware.</p>\n<p><strong>NOTE</strong>:  we don't know the frequency intervals used in the SETI comp so don't be gready in your conclusions.</p>\n<p>Now i wonder why the hosts didn't give us the data as signal in the SETI comp (I think it would have been much more interesting). </p>\n<p><em>Will someone of you be so brave to try transfer-learning or something fancy? ahahah</em></p>",
  "messages": [
    {
      "id": "1468523",
      "postDate": "08/12/2021 13:03:27",
      "content": "<p>I was playing with the data while I noticed that if we set:<br>\n<code>Spectrogram.CQT1992v2(sr=2048, fmin=20, fmax=1024, hop_length=5, bins_per_octave=48, pad_mode='constant')</code><br>\nwe can get a spectrogram that has  <code>freq_bins=273</code> amd <code>time_steps=820</code> (819 + 1 for appox) that is the <strong>same spectrum as the SETI challenge</strong>.</p>\n<p>There is <strong>one big downside</strong> CQT is quite expensive at this resolution. 3h/epoch regardless of GPU/CPU with my hardware.</p>\n<p><strong>NOTE</strong>:  we don't know the frequency intervals used in the SETI comp so don't be gready in your conclusions.</p>\n<p>Now i wonder why the hosts didn't give us the data as signal in the SETI comp (I think it would have been much more interesting). </p>\n<p><em>Will someone of you be so brave to try transfer-learning or something fancy? ahahah</em></p>",
      "rawMarkdown": "I was playing with the data while I noticed that if we set:\n`Spectrogram.CQT1992v2(sr=2048, fmin=20, fmax=1024, hop_length=5, bins_per_octave=48, pad_mode='constant')`\nwe can get a spectrogram that has  `freq_bins=273` amd `time_steps=820` (819 + 1 for appox) that is the **same spectrum as the SETI challenge**.\n\nThere is **one big downside** CQT is quite expensive at this resolution. 3h/epoch regardless of GPU/CPU with my hardware.\n\n**NOTE**:  we don't know the frequency intervals used in the SETI comp so don't be gready in your conclusions.\n\nNow i wonder why the hosts didn't give us the data as signal in the SETI comp (I think it would have been much more interesting). \n\n*Will someone of you be so brave to try transfer-learning or something fancy? ahahah*",
      "votes": null
    },
    {
      "id": "1471800",
      "postDate": "08/14/2021 13:24:31",
      "content": "<p>SETI data is FFT, not QTransform.</p>",
      "rawMarkdown": "SETI data is FFT, not QTransform.",
      "votes": null
    },
    {
      "id": "1489174",
      "postDate": "08/24/2021 18:54:11",
      "content": "<p>IMHO they are different competition, SETI is more like a pure image competition, but here you can use your singal processing knowledge. BTW I'm curious how's your result for such big CQT image size, how much gain did you get? Thanks. </p>",
      "rawMarkdown": "IMHO they are different competition, SETI is more like a pure image competition, but here you can use your singal processing knowledge. BTW I'm curious how's your result for such big CQT image size, how much gain did you get? Thanks.",
      "votes": null
    },
    {
      "id": "1561189",
      "postDate": "10/27/2021 12:15:40",
      "content": "<p>Hey All,</p>\n<p>Thank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey <a href=\"https://forms.gle/QP9L16niPexozyhu5\" target=\"_blank\">https://forms.gle/QP9L16niPexozyhu5</a>.</p>\n<p>Thank you all,</p>\n<p>Regards,<br>\nChris</p>",
      "rawMarkdown": "Hey All,\n\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.\n\nThank you all,\n\nRegards,\nChris",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1471800,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "08/14/2021 13:24:31",
      "content": "<p>SETI data is FFT, not QTransform.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1489174,
      "author_name": "superchenhao",
      "author_url": "",
      "post_date": "08/24/2021 18:54:11",
      "content": "<p>IMHO they are different competition, SETI is more like a pure image competition, but here you can use your singal processing knowledge. BTW I'm curious how's your result for such big CQT image size, how much gain did you get? Thanks. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1561189,
      "author_name": "zerafachris",
      "author_url": "",
      "post_date": "10/27/2021 12:15:40",
      "content": "<p>Hey All,</p>\n<p>Thank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey <a href=\"https://forms.gle/QP9L16niPexozyhu5\" target=\"_blank\">https://forms.gle/QP9L16niPexozyhu5</a>.</p>\n<p>Thank you all,</p>\n<p>Regards,<br>\nChris</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1468523": "I was playing with the data while I noticed that if we set:\n`Spectrogram.CQT1992v2(sr=2048, fmin=20, fmax=1024, hop_length=5, bins_per_octave=48, pad_mode='constant')`\nwe can get a spectrogram that has  `freq_bins=273` amd `time_steps=820` (819 + 1 for appox) that is the **same spectrum as the SETI challenge**.\n\nThere is **one big downside** CQT is quite expensive at this resolution. 3h/epoch regardless of GPU/CPU with my hardware.\n\n**NOTE**:  we don't know the frequency intervals used in the SETI comp so don't be gready in your conclusions.\n\nNow i wonder why the hosts didn't give us the data as signal in the SETI comp (I think it would have been much more interesting). \n\n*Will someone of you be so brave to try transfer-learning or something fancy? ahahah*",
    "1471800": "SETI data is FFT, not QTransform.",
    "1489174": "IMHO they are different competition, SETI is more like a pure image competition, but here you can use your singal processing knowledge. BTW I'm curious how's your result for such big CQT image size, how much gain did you get? Thanks.",
    "1561189": "Hey All,\n\nThank you all for taking part in our competition. The participation has been overwhelmingly positive. We are currently conducting a survey to gauge the demographic and outreach achieved. Kindly spare 2min and fill in this survey https://forms.gle/QP9L16niPexozyhu5.\n\nThank you all,\n\nRegards,\nChris"
  },
  "source": "meta"
}