Swimmer963 (Miranda Dixon-Luinenburg) | v1.11.0Sep 22nd 2020 | (-1) checked citation, is copied | ||
Swimmer963 (Miranda Dixon-Luinenburg) | v1.10.0Sep 16th 2020 | (+83/-9) | ||
Zack_M_Davis | v1.9.0Nov 17th 2009 | (-21) byline removal | ||
Vladimir_Nesov | v1.8.0Nov 8th 2009 | (+29) /* See also */ | ||
PeerInfinity | v1.7.0Sep 28th 2009 | |||
Vladimir_Nesov | v1.6.0Jun 26th 2009 | (+20/-27) | ||
Eliezer Yudkowsky | v1.5.0Jun 4th 2009 | (+578/-120) fleshed out | ||
Vladimir_Nesov | v1.4.0May 29th 2009 | (+19/-34) | ||
Vladimir_Nesov | v1.3.0May 29th 2009 | (+20/-21) /* See Also */ | ||
Vladimir_Nesov | v1.2.0May 29th 2009 | (+17) /* See Also */ |
"Inductive bias"bias refers to your suspicion that if the sun has risen for the last billion days in a row, then it may rise tomorrow as well. Since it is logically possible that the laws of physics will arbitrarily cease to work and that the sun will *not*not rise tomorrow, coming to this conclusion requires an inductively biased prior.
In"Inductive bias" refers to your suspicion that if the sun has risen for the last billion days in a row, then it may rise tomorrow as well. Since it is Bayesianlogically possible framework, inductivethat the laws of physics will arbitrarily cease to work and that the sun will *not* rise tomorrow, coming to this conclusion requires an inductively biased prior.
This sort of bias is encoded innot a bad thing - without "inductive bias" you can't draw any conclusion at all from the prior distribution.data. It's just a different technical meaning attached to the same word.