A General Lower Bound on the Number of Examples Needed for Learning

@inproceedings{Ehrenfeucht1988AGL,
  title={A General Lower Bound on the Number of Examples Needed for Learning},
  author={Andrzej Ehrenfeucht and David Haussler and Michael Kearns and Leslie G. Valiant},
  booktitle={COLT},
  year={1988}
}
We prove a lower bound of ( ln + VCdim(C) ) on the number of random examples required for distribution-free learning of a concept class C, where VCdim(C) is the Vapnik-Chervonenkis dimension and and are the accuracy and con dence parameters. This improves the previous best lower bound of ( ln + VCdim(C)), and comes close to the known general upper bound of… CONTINUE READING