The agreement property for independently trained classifiers
Statement
Let be a classifier family and let be a distribution. For independent train sets , let be classifiers trained on , respectively. The test accuracy is the probability that , while the agreement property. For certain classifier families and distributions , these probabilities are approximately equal:
Moreover, with high probability over the trained classifiers , the corresponding probabilities over are approximately equal. This is proposed as an instantiation of the paper's indistinguishability conjecture; it is provable for 1-nearest neighbors in some settings, but the general claim's resolution is not specified.
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.