neurodata
implement honest trees
An example of an honest tree is one where the tree is grown using one subsample, while the predictions at the leaves of the tree are estimated using a different subsample.
@j1c Ran an experiment showing that we don't need honest trees? @j1c please comment.
The experiment proposed in Biau (2016) was an example where RF fails, and I showed that RerF does not fail. But this doesn't necessarily prove that RerF is universally consistent, so honesty might still be worth implementing. Thoughts @jovo?
not worth implementing yet. check the RerF repo. i think we'll prove that we don't need honesty soon.
On Wed, Sep 5, 2018 at 3:00 PM Jaewon Chung [email protected] wrote:
The experiment proposed in Biau (2016) was an example where RF fails, and I showed that RerF does not fail. But this doesn't necessarily prove that RerF is universally consistent, so honesty might still be worth implementing. Thoughts @jovo https://github.com/jovo?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/neurodata/R-RerF/issues/36#issuecomment-418843092, or mute the thread https://github.com/notifications/unsubscribe-auth/AACjcjMdtAMQpfwzJIUS-g-tdbhGv0ENks5uYB8HgaJpZM4Uqigj .
-- the glass is all full: half water, half air. neurodata.io