{
  "id": 572327,
  "title": "Welcome to the Yale/UNC-Chapel Hill – Geophysical Waveform Inversion competition!",
  "url": "/competitions/waveform-inversion/discussion/572327",
  "author_name": "",
  "post_date": "2025-04-08T23:33:24.241068500Z",
  "votes": 15,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>Welcome to the <strong>Yale/UNC-Chapel Hill – Geophysical Waveform Inversion</strong> competition!</p>\n<p>We’re thrilled to bring you this unique challenge that bridges geophysics and machine learning. The goal of this competition is to reconstruct subsurface velocity maps from simulated seismic waveform data — a task with major implications for energy, environment, and geoscience research.</p>\n<p>Whether you’re a deep learning veteran or just curious about scientific ML, we hope you’ll find this both fun and meaningful. You’ll be working with waveform data and trying to “see” what lies underground — using only the ripples on the surface. Sounds cool? We think so too.</p>\n<p>We’ll be monitoring the forums closely and are excited to hear your ideas, feedback, and questions. Feel free to share approaches, issues, or just say hi 👋.</p>\n<p>Let’s build something great together!</p>\n<p>Best,<br>\nThe Waveform Inversion Team</p>",
  "messages": [
    {
      "id": "3174278",
      "postDate": "04/08/2025 23:33:24",
      "content": "<p>Hi everyone,</p>\n<p>Welcome to the <strong>Yale/UNC-Chapel Hill – Geophysical Waveform Inversion</strong> competition!</p>\n<p>We’re thrilled to bring you this unique challenge that bridges geophysics and machine learning. The goal of this competition is to reconstruct subsurface velocity maps from simulated seismic waveform data — a task with major implications for energy, environment, and geoscience research.</p>\n<p>Whether you’re a deep learning veteran or just curious about scientific ML, we hope you’ll find this both fun and meaningful. You’ll be working with waveform data and trying to “see” what lies underground — using only the ripples on the surface. Sounds cool? We think so too.</p>\n<p>We’ll be monitoring the forums closely and are excited to hear your ideas, feedback, and questions. Feel free to share approaches, issues, or just say hi 👋.</p>\n<p>Let’s build something great together!</p>\n<p>Best,<br>\nThe Waveform Inversion Team</p>",
      "rawMarkdown": "Hi everyone,\n\nWelcome to the **Yale/UNC-Chapel Hill – Geophysical Waveform Inversion** competition!\n\nWe’re thrilled to bring you this unique challenge that bridges geophysics and machine learning. The goal of this competition is to reconstruct subsurface velocity maps from simulated seismic waveform data — a task with major implications for energy, environment, and geoscience research.\n\nWhether you’re a deep learning veteran or just curious about scientific ML, we hope you’ll find this both fun and meaningful. You’ll be working with waveform data and trying to “see” what lies underground — using only the ripples on the surface. Sounds cool? We think so too.\n\nWe’ll be monitoring the forums closely and are excited to hear your ideas, feedback, and questions. Feel free to share approaches, issues, or just say hi 👋.\n\nLet’s build something great together!\n\nBest,\nThe Waveform Inversion Team",
      "votes": null
    },
    {
      "id": "3175602",
      "postDate": "04/10/2025 11:11:52",
      "content": "<p>Nice one indeed </p>",
      "rawMarkdown": "Nice one indeed",
      "votes": null
    },
    {
      "id": "3176003",
      "postDate": "04/10/2025 19:59:43",
      "content": "<p>Yep. It sounds cool! and Fun!</p>",
      "rawMarkdown": "Yep. It sounds cool! and Fun!",
      "votes": null
    },
    {
      "id": "3183176",
      "postDate": "04/20/2025 13:57:15",
      "content": "<p>Really having fun with this one - finally get to use some of my college classes from 55 years ago.</p>\n<p>But - not sure you could have made downloading the full dataset any harder.</p>\n<p>Google drive a bit of a pain.   Splitting files into 3 per folder I guess makes a little sense, but still a bit of work to get the additional files into the kaggle download structure.</p>\n<p>But my biggest issue - having at least two ( I have not downloaded all the files yet) different file structures makes no sense.  Hopefully I only need to end up writing code to get two different structures into the kaggle folder setup.</p>\n<p>So please - next time - think about saving lots of folks lots of time - pretty sure anyone who wants to be in the medals needs to use the full data set.   Its looking like I am going to spend close to 12 hours getting the full set.</p>",
      "rawMarkdown": "Really having fun with this one - finally get to use some of my college classes from 55 years ago.\n\nBut - not sure you could have made downloading the full dataset any harder.\n\nGoogle drive a bit of a pain.   Splitting files into 3 per folder I guess makes a little sense, but still a bit of work to get the additional files into the kaggle download structure.\n\nBut my biggest issue - having at least two ( I have not downloaded all the files yet) different file structures makes no sense.  Hopefully I only need to end up writing code to get two different structures into the kaggle folder setup.\n\nSo please - next time - think about saving lots of folks lots of time - pretty sure anyone who wants to be in the medals needs to use the full data set.   Its looking like I am going to spend close to 12 hours getting the full set.",
      "votes": null
    },
    {
      "id": "3183286",
      "postDate": "04/20/2025 17:18:22",
      "content": "<p>We’re very glad to hear that you’re enjoying the competition, <a href=\"https://www.kaggle.com/pcjimmmy\" target=\"_blank\">@pcjimmmy</a>! This is exactly what we hoped to achieve by creating and sharing this Kaggle Competition with the broader Data Science / Machine Learning community. One of our main goals is to highlight the exciting potential of AI for Science, especially in wave imaging applications.<br>\nAgain, thank you for your thoughtful suggestions regarding the dataset and download process. We truly appreciate your feedback and will take it into serious consideration for future research and data releases.<br>\nFor this competition, a duplicate of the OpenFWI dataset has been uploaded to Hugging Face. You can find it here: <a href=\"https://huggingface.co/datasets/ashynf/OpenFWI\" target=\"_blank\">https://huggingface.co/datasets/ashynf/OpenFWI</a><br>\nAlso, we’d like to highlight this excellent summary post shared by <a href=\"https://www.kaggle.com/seshurajup\" target=\"_blank\">@seshurajup</a> (thank you!): <a href=\"https://www.kaggle.com/competitions/waveform-inversion/discussion/572434\" target=\"_blank\">https://www.kaggle.com/competitions/waveform-inversion/discussion/572434</a><br>\nThanks again to all participants for your engagement and contributions! We’re excited to learn about your creative ML models! <br>\n-The Waveform Inversion Team</p>",
      "rawMarkdown": "We’re very glad to hear that you’re enjoying the competition, @pcjimmmy! This is exactly what we hoped to achieve by creating and sharing this Kaggle Competition with the broader Data Science / Machine Learning community. One of our main goals is to highlight the exciting potential of AI for Science, especially in wave imaging applications.\n\nAgain, thank you for your thoughtful suggestions regarding the dataset and download process. We truly appreciate your feedback and will take it into serious consideration for future research and data releases.\n\nFor this competition, a duplicate of the OpenFWI dataset has been uploaded to Hugging Face. You can find it here: [https://huggingface.co/datasets/ashynf/OpenFWI](https://huggingface.co/datasets/ashynf/OpenFWI)\n\nAlso, we’d like to highlight this excellent summary post shared by @seshurajup (thank you!): [https://www.kaggle.com/competitions/waveform-inversion/discussion/572434](https://www.kaggle.com/competitions/waveform-inversion/discussion/572434)\n\nThanks again to all participants for your engagement and contributions! We’re excited to learn about your creative ML models! \n\n\n-The Waveform Inversion Team",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3175602,
      "author_name": "phokamatete",
      "author_url": "",
      "post_date": "04/10/2025 11:11:52",
      "content": "<p>Nice one indeed </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3176003,
      "author_name": "harharhari",
      "author_url": "",
      "post_date": "04/10/2025 19:59:43",
      "content": "<p>Yep. It sounds cool! and Fun!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3183176,
      "author_name": "pcjimmmy",
      "author_url": "",
      "post_date": "04/20/2025 13:57:15",
      "content": "<p>Really having fun with this one - finally get to use some of my college classes from 55 years ago.</p>\n<p>But - not sure you could have made downloading the full dataset any harder.</p>\n<p>Google drive a bit of a pain.   Splitting files into 3 per folder I guess makes a little sense, but still a bit of work to get the additional files into the kaggle download structure.</p>\n<p>But my biggest issue - having at least two ( I have not downloaded all the files yet) different file structures makes no sense.  Hopefully I only need to end up writing code to get two different structures into the kaggle folder setup.</p>\n<p>So please - next time - think about saving lots of folks lots of time - pretty sure anyone who wants to be in the medals needs to use the full data set.   Its looking like I am going to spend close to 12 hours getting the full set.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3183286,
          "author_name": "youzuolin",
          "author_url": "",
          "post_date": "04/20/2025 17:18:22",
          "content": "<p>We’re very glad to hear that you’re enjoying the competition, <a href=\"https://www.kaggle.com/pcjimmmy\" target=\"_blank\">@pcjimmmy</a>! This is exactly what we hoped to achieve by creating and sharing this Kaggle Competition with the broader Data Science / Machine Learning community. One of our main goals is to highlight the exciting potential of AI for Science, especially in wave imaging applications.<br>\nAgain, thank you for your thoughtful suggestions regarding the dataset and download process. We truly appreciate your feedback and will take it into serious consideration for future research and data releases.<br>\nFor this competition, a duplicate of the OpenFWI dataset has been uploaded to Hugging Face. You can find it here: <a href=\"https://huggingface.co/datasets/ashynf/OpenFWI\" target=\"_blank\">https://huggingface.co/datasets/ashynf/OpenFWI</a><br>\nAlso, we’d like to highlight this excellent summary post shared by <a href=\"https://www.kaggle.com/seshurajup\" target=\"_blank\">@seshurajup</a> (thank you!): <a href=\"https://www.kaggle.com/competitions/waveform-inversion/discussion/572434\" target=\"_blank\">https://www.kaggle.com/competitions/waveform-inversion/discussion/572434</a><br>\nThanks again to all participants for your engagement and contributions! We’re excited to learn about your creative ML models! <br>\n-The Waveform Inversion Team</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3174278": "Hi everyone,\n\nWelcome to the **Yale/UNC-Chapel Hill – Geophysical Waveform Inversion** competition!\n\nWe’re thrilled to bring you this unique challenge that bridges geophysics and machine learning. The goal of this competition is to reconstruct subsurface velocity maps from simulated seismic waveform data — a task with major implications for energy, environment, and geoscience research.\n\nWhether you’re a deep learning veteran or just curious about scientific ML, we hope you’ll find this both fun and meaningful. You’ll be working with waveform data and trying to “see” what lies underground — using only the ripples on the surface. Sounds cool? We think so too.\n\nWe’ll be monitoring the forums closely and are excited to hear your ideas, feedback, and questions. Feel free to share approaches, issues, or just say hi 👋.\n\nLet’s build something great together!\n\nBest,\nThe Waveform Inversion Team",
    "3175602": "Nice one indeed",
    "3176003": "Yep. It sounds cool! and Fun!",
    "3183176": "Really having fun with this one - finally get to use some of my college classes from 55 years ago.\n\nBut - not sure you could have made downloading the full dataset any harder.\n\nGoogle drive a bit of a pain.   Splitting files into 3 per folder I guess makes a little sense, but still a bit of work to get the additional files into the kaggle download structure.\n\nBut my biggest issue - having at least two ( I have not downloaded all the files yet) different file structures makes no sense.  Hopefully I only need to end up writing code to get two different structures into the kaggle folder setup.\n\nSo please - next time - think about saving lots of folks lots of time - pretty sure anyone who wants to be in the medals needs to use the full data set.   Its looking like I am going to spend close to 12 hours getting the full set.",
    "3183286": "We’re very glad to hear that you’re enjoying the competition, @pcjimmmy! This is exactly what we hoped to achieve by creating and sharing this Kaggle Competition with the broader Data Science / Machine Learning community. One of our main goals is to highlight the exciting potential of AI for Science, especially in wave imaging applications.\n\nAgain, thank you for your thoughtful suggestions regarding the dataset and download process. We truly appreciate your feedback and will take it into serious consideration for future research and data releases.\n\nFor this competition, a duplicate of the OpenFWI dataset has been uploaded to Hugging Face. You can find it here: [https://huggingface.co/datasets/ashynf/OpenFWI](https://huggingface.co/datasets/ashynf/OpenFWI)\n\nAlso, we’d like to highlight this excellent summary post shared by @seshurajup (thank you!): [https://www.kaggle.com/competitions/waveform-inversion/discussion/572434](https://www.kaggle.com/competitions/waveform-inversion/discussion/572434)\n\nThanks again to all participants for your engagement and contributions! We’re excited to learn about your creative ML models! \n\n\n-The Waveform Inversion Team"
  },
  "source": "meta"
}