Avrim Blum
- 1.N. Alon and J. Spencer. The ProbabilisticMethod. Wiley, 1991.Google Scholar
- 2.M. Anthony and S. B. Holden Cross-Validation for Binary Classification by Real-Valued Functions: Theoretical Analysis In Proc. Eleventh Annual Conference on Computational Learning Theory, 1998. Google ScholarDigital Library
- 3.L. Brieman. Bagging predictors. Machine Learning, 24(2): 123-140, 1!)96. Google ScholarDigital Library
- 4.L. Devroye, L. Gyrofi, and G. Lugosi. A Probabilistic Theory of Pattern Recognition. Springer-Verlag, 1996.Google ScholarCross Ref
- 5.L. Devroye and T. Wagner. Distribution-free performance bounds for potential function rules. IEEE Transactions on Pattern Analysis and Machine Intelligence, IT-25(5):601-604, 1979. Google ScholarDigital Library
- 6.Y. Freund and R. Shapire. Discussion of the paper "Arcing classifiers" by Leo Breiman. Annals of Statistics, 26(3): 824-832, 1998.Google Scholar
- 7.D.P. Helmbold and M.K. Warmuth. On Weak Learning. JCSS, 50(3): 551-573, 1995. Google ScholarDigital Library
- 8.M. J. Kearns and D. Ron. Algorithmic stability and sanity-check bounds for leave-one-out crossvalidation. In Proc. Tenth Annual Conference on Computational Learning Theory, 1997. Google ScholarDigital Library
- 9.N. Littlestone. From on-line to batch learning. In Proceedings of the 2nd Annual Workshop on Computational Learning Theory, pp. 269-284, 1989. Google ScholarDigital Library
Index Terms
- Beating the hold-out: bounds for K-fold and progressive cross-validation
Recommendations
Aggregated hold-out
Aggregated hold-out (agghoo) is a method which averages learning rules selected by holdout (that is, cross-validation with a single split). We provide the first theoretical guarantees on agghoo, ensuring that it can be used safely: Agghoo performs at ...
Training Orchestral Conductors in Beating Time
SA '23: SIGGRAPH Asia 2023 Technical CommunicationsOrchestral conducting involves a rich vocabulary of gestures and so training conductors is challenging. We discuss how virtual reality and gesture detection could be used to aid this process. We describe our pilot interface for training conductors in ...
Beating the probabilistic lower bound on perfect hashing
SODA '21: Proceedings of the Thirty-Second Annual ACM-SIAM Symposium on Discrete AlgorithmsFor an integer q ⩾ 2, a perfect q-hash code C is a block code over [q] := {1,…,q} of length n in which every subset {c1, c2,…,cq} of q elements is separated, i.e., there exists i ∈ [n] such that {proji(c1),…,proji(cq)} = [q], where proji(cj) denotes the ...
Comments