Making the rounds again... I like the promise of making any ML algorithm probabilistic.
...Blackbox #MachineLearning models are now routinely used in high-risk settings, like medical diagnostics, which demand uncertainty quantification to avoid consequential model failures... #ConformalPrediction is a user-friendly paradigm for creating statistically rigorous uncertainty sets/intervals for the predictions of such models...
[1] https://arxiv.org/abs/2107.07511
[2] https://arxiv.org/abs/2106.06137
Kyle Taylor /
npub1mj…yuq8j
2024-07-03 21:42:03
Author Public Key
npub1mjq8tusfxfxatffmz8lfrm5gvdkzutjetl88spr576jy2jc2hjyq3yuq8jPublished at
2024-07-03 21:42:03Event JSON
{
"id": "6ea1fb197ca445d4d97d856d74e4964868e071f690142709747125cc2c44467a",
"pubkey": "dc8075f209324dd5a53b11fe91ee88636c2e2e595fce780474f6a4454b0abc88",
"created_at": 1720042923,
"kind": 1,
"tags": [
[
"t",
"machinelearning"
],
[
"t",
"conformalprediction"
],
[
"proxy",
"https://hostux.social/users/kta/statuses/112724733004174794",
"activitypub"
]
],
"content": "Making the rounds again... I like the promise of making any ML algorithm probabilistic.\n\n...Blackbox #MachineLearning models are now routinely used in high-risk settings, like medical diagnostics, which demand uncertainty quantification to avoid consequential model failures... #ConformalPrediction is a user-friendly paradigm for creating statistically rigorous uncertainty sets/intervals for the predictions of such models... \n\n[1] https://arxiv.org/abs/2107.07511\n[2] https://arxiv.org/abs/2106.06137",
"sig": "8f874cffc0c7624fd51f7391cdb42334f6ab072eb8646c7987266e29fb322473818c933ac293bade5d66f1afd4ec15bd065a353efa4acb74bf654e5cf0c90b8c"
}