\n
![\"quaternion](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4310004%2F4a77f2c11f1929e21f3758529c772e41%2Fim1.png?generation=1756861811982691&alt=media\")
I want to thank my teammates @mahdiseddigh, @shanzhong8, @guoooooooss and @leonshangguan. You all put in your best effort to manage your time for this tough competition. Even though we didn’t perform well last month, at least we got the chance to review our college physics course knowledge.
\nWe observed that different behaviors lead to different results depending on the phase. This suggests that carefully manipulating the signal during a significant phase could be highly beneficial, as the phase plays a critical role in determining the outcome.
\nQuaternion error:
\n
Gravity accel error:
\n![\"gravity](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4310004%2F082621d99a0023b4a7f644644d8e4884%2Fim2.png?generation=1756861828667195&alt=media\")
Then it turns out that we find a good library to handle this type of signal, using ATAR algorithm in spkit to process the signal in each phase, and boost cv and lb.
\nHere are the result using different parameter in ATAR on acceleration:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4310004%2F53bb2e5271afda37bff585cc6b6fb8d5%2FAtar.png?generation=1756862375147492&alt=media\")
We also try to use Faiss to retrieve similar signal and use them in the inference, just miss to use subject which is really important in this competition.
\nWill fill up more later.
", "messages": [ { "id": "3280646", "postDate": "09/03/2025 01:21:10", "content": "I want to thank my teammates @mahdiseddigh, @shanzhong8, @guoooooooss and @leonshangguan. You all put in your best effort to manage your time for this tough competition. Even though we didn’t perform well last month, at least we got the chance to review our college physics course knowledge.
\nWe observed that different behaviors lead to different results depending on the phase. This suggests that carefully manipulating the signal during a significant phase could be highly beneficial, as the phase plays a critical role in determining the outcome.
\nQuaternion error:
\n
Gravity accel error:
\n
Then it turns out that we find a good library to handle this type of signal, using ATAR algorithm in spkit to process the signal in each phase, and boost cv and lb.
\nHere are the result using different parameter in ATAR on acceleration:
\n
We also try to use Faiss to retrieve similar signal and use them in the inference, just miss to use subject which is really important in this competition.
\nWill fill up more later.
", "rawMarkdown": "### Acknowledgement\n\nI want to thank my teammates @mahdiseddigh, @shanzhong8, @guoooooooss and @leonshangguan. You all put in your best effort to manage your time for this tough competition. Even though we didn’t perform well last month, at least we got the chance to review our college physics course knowledge.\n\n\n### Solution\n\nWe observed that different behaviors lead to different results depending on the phase. This suggests that carefully manipulating the signal during a significant phase could be highly beneficial, as the phase plays a critical role in determining the outcome.\n\nQuaternion error: \n\n\nGravity accel error: \n\n\nThen it turns out that we find a good library to handle this type of signal, using ATAR algorithm in [spkit](https://spkit.github.io/modules/generated/spkit.eeg.ATAR.html#spkit.eeg.ATAR) to process the signal in each phase, and boost cv and lb. \n\nHere are the result using different parameter in ATAR on acceleration:\n\n\n\nWe also try to use [Faiss](https://github.com/facebookresearch/faiss) to retrieve similar signal and use them in the inference, just miss to use subject which is really important in this competition. \n\n\nWill fill up more later.", "votes": null }, { "id": "3280909", "postDate": "09/03/2025 10:53:56", "content": "A good team and good solutions and welcome, friend @tom99763
", "rawMarkdown": "A good team and good solutions and welcome, friend @tom99763", "votes": null }, { "id": "3293497", "postDate": "09/24/2025 02:36:35", "content": "Hoping to hear more detail~
", "rawMarkdown": "Hoping to hear more detail~", "votes": null } ], "comments": [ { "id": 3280909, "author_name": "muhammaddanyalmalik", "author_url": "", "post_date": "09/03/2025 10:53:56", "content": "A good team and good solutions and welcome, friend @tom99763
", "votes": null, "replies": [] }, { "id": 3293497, "author_name": "xukongji", "author_url": "", "post_date": "09/24/2025 02:36:35", "content": "Hoping to hear more detail~
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3280646": "### Acknowledgement\n\nI want to thank my teammates @mahdiseddigh, @shanzhong8, @guoooooooss and @leonshangguan. You all put in your best effort to manage your time for this tough competition. Even though we didn’t perform well last month, at least we got the chance to review our college physics course knowledge.\n\n\n### Solution\n\nWe observed that different behaviors lead to different results depending on the phase. This suggests that carefully manipulating the signal during a significant phase could be highly beneficial, as the phase plays a critical role in determining the outcome.\n\nQuaternion error: \n\n\nGravity accel error: \n\n\nThen it turns out that we find a good library to handle this type of signal, using ATAR algorithm in [spkit](https://spkit.github.io/modules/generated/spkit.eeg.ATAR.html#spkit.eeg.ATAR) to process the signal in each phase, and boost cv and lb. \n\nHere are the result using different parameter in ATAR on acceleration:\n\n\n\nWe also try to use [Faiss](https://github.com/facebookresearch/faiss) to retrieve similar signal and use them in the inference, just miss to use subject which is really important in this competition. \n\n\nWill fill up more later.", "3280909": "A good team and good solutions and welcome, friend @tom99763", "3293497": "Hoping to hear more detail~" }, "source": "existing_discussion", "complete": true, "competition_id": 102335, "competition_slug": "cmi-detect-behavior-with-sensor-data", "team_id": 13895665, "solution_url": "/competitions/cmi-detect-behavior-with-sensor-data/writeups/110th-place-solution" }