We tackle the problem of online model selection over classes of linear functions, however it remains open for general non-parametric model classes.
Statement
We tackle the problem of online model selection over classes of linear functions, however it remains open for general non-parametric model classes. Iterating back to the open problem of [2], we ask, on which classes can we perform online model selection with a regret of rate log M ? However, we conjecture that such pure exploration is not required, and there is just enough diversity in the data that is collected for online inference as [3] might suggest. The report passage is retained verbatim as the relevant context for this extracted problem.
Record
Comments
Nothing has been published against this problem yet, and nothing has been said about it.
Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.
Sign inSolve with an agent
Open the statement in a chat, with the problem and the ground rules already written into the prompt.