If you are new to CNN's, new to Kaggle, new to this competition, or you just want some ideas which kernel to check out next, here is a list 6 useful kernels:
\n\nSolution prepared by a Kaggle Grandmaster for the original MNIST competition now adapted for the Kannada dataset here.
Fascinating visualisations of the input data produced by unsupervised approaches especially T-SNE here.
Model with High Accuracy and a tutorial explaining CNN's here.
Approach using FastAI here.
Tutorial CNN notebook here.
Model exploring validation approaches, contains useful explanation of CNN's, explains each step of the process here.
What helped you to progress in this competition? Add a comment for kernels you found useful.
", "messages": [ { "id": "669232", "postDate": "11/09/2019 18:17:14", "content": "If you are new to CNN's, new to Kaggle, new to this competition, or you just want some ideas which kernel to check out next, here is a list 6 useful kernels:
\n\nSolution prepared by a Kaggle Grandmaster for the original MNIST competition now adapted for the Kannada dataset here.
Fascinating visualisations of the input data produced by unsupervised approaches especially T-SNE here.
Model with High Accuracy and a tutorial explaining CNN's here.
Approach using FastAI here.
Tutorial CNN notebook here.
Model exploring validation approaches, contains useful explanation of CNN's, explains each step of the process here.
What helped you to progress in this competition? Add a comment for kernels you found useful.
", "rawMarkdown": "If you are new to CNN's, new to Kaggle, new to this competition, or you just want some ideas which kernel to check out next, here is a list 6 useful kernels:\n\n1. Solution prepared by a Kaggle Grandmaster for the original MNIST competition now adapted for the Kannada dataset [here](https://www.kaggle.com/datahobbit/grandmaster-solution-adapted-from-mnist).\n\n2. Fascinating visualisations of the input data produced by unsupervised approaches especially T-SNE [here](https://www.kaggle.com/parulpandey/visualizing-kannada-mnist-with-t-sne).\n\n3. Model with High Accuracy and a tutorial explaining CNN's [here](https://www.kaggle.com/kenanajk/understanding-cnns-with-kannada-mnist).\n\n4. Approach using FastAI [here](https://www.kaggle.com/zinovadr/mnist-fastai-pytorch-0-98800-public-score).\n\n5. Tutorial CNN notebook [here](https://www.kaggle.com/shahules/indian-way-to-learn-cnn).\n\n6. Model exploring validation approaches, contains useful explanation of CNN's, explains each step of the process [here](https://www.kaggle.com/datahobbit/cnn-for-digit-recognition-in-the-kannada-script).\n\nWhat helped you to progress in this competition? Add a comment for kernels you found useful.", "votes": null }, { "id": "669261", "postDate": "11/09/2019 19:14:20", "content": "I used kernel one that you have listed for most of my learning. But only, by trying and failing has the knowledge gone in. I always have to learn the hard way unfortunatly
", "rawMarkdown": "I used kernel one that you have listed for most of my learning. But only, by trying and failing has the knowledge gone in. I always have to learn the hard way unfortunatly", "votes": null }, { "id": "669265", "postDate": "11/09/2019 19:26:35", "content": "That's the best way to learn a lot of the time. You're not the only one who has that experience! 🙂
", "rawMarkdown": "That's the best way to learn a lot of the time. You're not the only one who has that experience! 🙂", "votes": null }, { "id": "681508", "postDate": "11/26/2019 07:18:13", "content": "Great set of kernels
", "rawMarkdown": "Great set of kernels", "votes": null }, { "id": "682062", "postDate": "11/26/2019 21:35:19", "content": "I thought they were pretty good! Would be great to hear any other kernels that people found helpful.
", "rawMarkdown": "I thought they were pretty good! Would be great to hear any other kernels that people found helpful.", "votes": null }, { "id": "683973", "postDate": "11/29/2019 05:09:34", "content": "very fascinating
", "rawMarkdown": "very fascinating", "votes": null }, { "id": "684449", "postDate": "11/29/2019 19:13:27", "content": "Thanks!
", "rawMarkdown": "Thanks!", "votes": null }, { "id": "684793", "postDate": "11/30/2019 13:36:27", "content": "Great list of kernels, thanks for sharing Michael!
", "rawMarkdown": "Great list of kernels, thanks for sharing Michael!", "votes": null }, { "id": "688949", "postDate": "12/06/2019 09:22:38", "content": "Thanks - glad it was helpful.
", "rawMarkdown": "Thanks - glad it was helpful.", "votes": null } ], "comments": [ { "id": 669261, "author_name": "jakelj", "author_url": "", "post_date": "11/09/2019 19:14:20", "content": "I used kernel one that you have listed for most of my learning. But only, by trying and failing has the knowledge gone in. I always have to learn the hard way unfortunatly
", "votes": null, "replies": [ { "id": 669265, "author_name": "datahobbit", "author_url": "", "post_date": "11/09/2019 19:26:35", "content": "That's the best way to learn a lot of the time. You're not the only one who has that experience! 🙂
", "votes": null, "replies": [] } ] }, { "id": 681508, "author_name": "sachingoswami", "author_url": "", "post_date": "11/26/2019 07:18:13", "content": "Great set of kernels
", "votes": null, "replies": [ { "id": 682062, "author_name": "datahobbit", "author_url": "", "post_date": "11/26/2019 21:35:19", "content": "I thought they were pretty good! Would be great to hear any other kernels that people found helpful.
", "votes": null, "replies": [] } ] }, { "id": 683973, "author_name": "saurav21", "author_url": "", "post_date": "11/29/2019 05:09:34", "content": "very fascinating
", "votes": null, "replies": [ { "id": 684449, "author_name": "datahobbit", "author_url": "", "post_date": "11/29/2019 19:13:27", "content": "Thanks!
", "votes": null, "replies": [] } ] }, { "id": 684793, "author_name": "stefanoleone992", "author_url": "", "post_date": "11/30/2019 13:36:27", "content": "Great list of kernels, thanks for sharing Michael!
", "votes": null, "replies": [ { "id": 688949, "author_name": "datahobbit", "author_url": "", "post_date": "12/06/2019 09:22:38", "content": "Thanks - glad it was helpful.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "669232": "If you are new to CNN's, new to Kaggle, new to this competition, or you just want some ideas which kernel to check out next, here is a list 6 useful kernels:\n\n1. Solution prepared by a Kaggle Grandmaster for the original MNIST competition now adapted for the Kannada dataset [here](https://www.kaggle.com/datahobbit/grandmaster-solution-adapted-from-mnist).\n\n2. Fascinating visualisations of the input data produced by unsupervised approaches especially T-SNE [here](https://www.kaggle.com/parulpandey/visualizing-kannada-mnist-with-t-sne).\n\n3. Model with High Accuracy and a tutorial explaining CNN's [here](https://www.kaggle.com/kenanajk/understanding-cnns-with-kannada-mnist).\n\n4. Approach using FastAI [here](https://www.kaggle.com/zinovadr/mnist-fastai-pytorch-0-98800-public-score).\n\n5. Tutorial CNN notebook [here](https://www.kaggle.com/shahules/indian-way-to-learn-cnn).\n\n6. Model exploring validation approaches, contains useful explanation of CNN's, explains each step of the process [here](https://www.kaggle.com/datahobbit/cnn-for-digit-recognition-in-the-kannada-script).\n\nWhat helped you to progress in this competition? Add a comment for kernels you found useful.", "669261": "I used kernel one that you have listed for most of my learning. But only, by trying and failing has the knowledge gone in. I always have to learn the hard way unfortunatly", "669265": "That's the best way to learn a lot of the time. You're not the only one who has that experience! 🙂", "681508": "Great set of kernels", "682062": "I thought they were pretty good! Would be great to hear any other kernels that people found helpful.", "683973": "very fascinating", "684449": "Thanks!", "684793": "Great list of kernels, thanks for sharing Michael!", "688949": "Thanks - glad it was helpful." }, "source": "meta" }