Timeline for The Two Cultures: statistics vs. machine learning?
Current License: CC BY-SA 2.5
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 11, 2011 at 2:25 | comment | added | Muhammad Alkarouri | As a ML practitioner, I don't recognise the picture you are painting. The ML literature is almost all about variations of regularisation, MDL, Bayesian, SRM and other approaches of controlling the complexity of the model. From where I sit, it seems that stat's methods of controlling complexity are less structured, but that is bias for you. | |
| May 29, 2011 at 16:03 | comment | added | probabilityislogic | your comment about The "best" model will be choisen by the ML practitioner... applies equally well to mainstream statistics as well. For in most model selection procedures, one simply conditions on the final model as if no search of the model space had been done (given that model averaging is fairly new). So I don't think you can use that as a "club" to beat the ML practitioner with, so to speak. | |
| Aug 9, 2010 at 13:56 | history | edited | Thylacoleo | CC BY-SA 2.5 |
added 72 characters in body
|
| Aug 9, 2010 at 13:51 | history | answered | Thylacoleo | CC BY-SA 2.5 |