Abstract
In this short note, we show that the VC-dimension of the class of k-vertex polytopes in Rd is at most 8d2k log2k, answering an old question of Long and Warmuth. We also show that it is at least 1/3kd.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
A. Blumer, A. Ehrenfeucht, D. Haussler and M. K. Warmuth: Learnability and the Vapnik-Chervonenkis dimension, J. ACM 36 (1989), 929–965.
M. Csikós, A. Kupavskii and N. Mustafa Optimal bounds on the VC-dimension, Journal of Machine Learning Research 20 (2019), 81.1–81.8.
P. M. Long and M. K. Warmuth: Composite Geometric Concepts and Polynomial Predictability, COLT (1990), 273–287.
J. Matoušek: Lectures on Discrete Geometry, Springer, Vol. 212 (2002), New York.
J. W. Milnor On the Betti numbers of real algebraic varieties, Proc. Amer. Math. Soc. 15 (1964), 275–280.
O. A. Oleinik and I. B. Petrrovskii: On the topology of real algebraic surfaces, Izv. Akad. Nauk SSSR 13 (1949), 389–402 (in Russian).
R. Thom: On the homology of real algebraic varieties, in: S. S. Cairns, editor, Differential and Combinatorial Topology, Princeton Univ. Press, 1965 (in French).
V. N. Vapnik and A. Ya. Chervonenkis: On the uniform convergence of relative frequenciesof events to their probabilities, Theory of Probability and its Applications 16 (1971), 264–280.
Acknowledgements
We thank Lee-Ad Gottlieb, Aryeh Kontorovich and Gabriel Nivasch for sharing the question and for several enlightening discussions. We also thank the referee for his useful comments, which, in particular, significantly improved the presentation of the lower bound.
Author information
Authors and Affiliations
Corresponding author
Additional information
The author acknowledges the financial support from the Ministry of Education and Science of the Russian Federation in the framework of MegaGrant no 075-15-2019-1926.
