I hate it when most of the top solutions compete by the number of models in their stacks or how heavy their encoders are. But this competition is so different.
\n1st, 5th and 6th places use non-machine learning approaches
\n1st place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375910
\n5th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/376022
\n6th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375923
9th place solution is elegant and effective https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375897
\n12th place solution is relatively cheap in terms of compute (single convnext tiny/small) - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375961
\n22nd place gives an interesting insight into the importance of signal depth for CNN models https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375927
\n@rodrigotenorio you and the team must be pleased with the results. How do the top results compare to those of traditional methods?
", "messages": [ { "id": 2085543, "postDate": "2023-01-04T08:14:56.417Z", "content": "I hate it when most of the top solutions compete by the number of models in their stacks or how heavy their encoders are. But this competition is so different.
\n1st, 5th and 6th places use non-machine learning approaches
\n1st place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375910
\n5th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/376022
\n6th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375923
9th place solution is elegant and effective https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375897
\n12th place solution is relatively cheap in terms of compute (single convnext tiny/small) - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375961
\n22nd place gives an interesting insight into the importance of signal depth for CNN models https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375927
\n@rodrigotenorio you and the team must be pleased with the results. How do the top results compare to those of traditional methods?
", "rawMarkdown": "I hate it when most of the top solutions compete by the number of models in their stacks or how heavy their encoders are. But this competition is so different. \n1st, 5th and 6th places use non-machine learning approaches\n1st place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375910\n5th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/376022\n6th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375923\n\n9th place solution is elegant and effective https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375897\n\n12th place solution is relatively cheap in terms of compute (single convnext tiny/small) - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375961\n\n22nd place gives an interesting insight into the importance of signal depth for CNN models https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375927\n\n@rodrigotenorio you and the team must be pleased with the results. How do the top results compare to those of traditional methods?\n", "votes": 7 } ], "comments": [], "raw_markdown_by_id": { "2085543": "I hate it when most of the top solutions compete by the number of models in their stacks or how heavy their encoders are. But this competition is so different. \n1st, 5th and 6th places use non-machine learning approaches\n1st place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375910\n5th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/376022\n6th place - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375923\n\n9th place solution is elegant and effective https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375897\n\n12th place solution is relatively cheap in terms of compute (single convnext tiny/small) - https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375961\n\n22nd place gives an interesting insight into the importance of signal depth for CNN models https://www.kaggle.com/competitions/g2net-detecting-continuous-gravitational-waves/discussion/375927\n\n@rodrigotenorio you and the team must be pleased with the results. How do the top results compare to those of traditional methods?\n" } }