Hey,\n I tried a lot of methods and complicated models even hybriding Convolutional Neural Network + XGBoost + Gaussian Naive Bayers, but couldn't cross the accuracy of 0.9900 with any type of image augmentation or layer depth, So I would like to request the 10% to kindly share their view on how to solve a general problem like this and what is the small things we are (remaining participant) neglecting or overlooking generally, I am looking for a general answers which will help any newbie ML enthusiast to perform better in next.
", "messages": [ { "id": "703787", "postDate": "12/26/2019 16:09:24", "content": "Hey,\n I tried a lot of methods and complicated models even hybriding Convolutional Neural Network + XGBoost + Gaussian Naive Bayers, but couldn't cross the accuracy of 0.9900 with any type of image augmentation or layer depth, So I would like to request the 10% to kindly share their view on how to solve a general problem like this and what is the small things we are (remaining participant) neglecting or overlooking generally, I am looking for a general answers which will help any newbie ML enthusiast to perform better in next.
", "rawMarkdown": "Hey,\n I tried a lot of methods and complicated models even [hybriding Convolutional Neural Network + XGBoost + Gaussian Naive Bayers](https://www.kaggle.com/chekoduadarsh/hybrid-cnn-xgboost-gnb-accuracy-99), but couldn't cross the accuracy of 0.9900 with any type of image augmentation or layer depth, So I would like to request the 10% to kindly share their view on how to solve a general problem like this and what is the small things we are (remaining participant) neglecting or overlooking generally, I am looking for a general answers which will help any newbie ML enthusiast to perform better in next.", "votes": null }, { "id": "703814", "postDate": "12/26/2019 17:02:25", "content": "You can check out my solution for theory here :- Solution Explained Top 1%
\n\nCheckout my Kernel at :- Kernel
", "rawMarkdown": "You can check out my solution for theory here :- [Solution Explained Top 1%](https://www.kaggle.com/c/Kannada-MNIST/discussion/122158)\n\nCheckout my Kernel at :- [Kernel](https://www.kaggle.com/namanj27/11th-place-solution-code)", "votes": null } ], "comments": [ { "id": 703814, "author_name": "namanj27", "author_url": "", "post_date": "12/26/2019 17:02:25", "content": "You can check out my solution for theory here :- Solution Explained Top 1%
\n\nCheckout my Kernel at :- Kernel
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "703787": "Hey,\n I tried a lot of methods and complicated models even [hybriding Convolutional Neural Network + XGBoost + Gaussian Naive Bayers](https://www.kaggle.com/chekoduadarsh/hybrid-cnn-xgboost-gnb-accuracy-99), but couldn't cross the accuracy of 0.9900 with any type of image augmentation or layer depth, So I would like to request the 10% to kindly share their view on how to solve a general problem like this and what is the small things we are (remaining participant) neglecting or overlooking generally, I am looking for a general answers which will help any newbie ML enthusiast to perform better in next.", "703814": "You can check out my solution for theory here :- [Solution Explained Top 1%](https://www.kaggle.com/c/Kannada-MNIST/discussion/122158)\n\nCheckout my Kernel at :- [Kernel](https://www.kaggle.com/namanj27/11th-place-solution-code)" }, "source": "meta" }