A. Smola, N. Murata, B. Sch\"olkopf, and K.-R. M\"uller (1998)
Asymptotically Optimal Choice of $\varepsilon$-Loss for Support Vector Machines
In: Proceedings of ICANN'98, ed. by L. Niklasson and M. Bod\'en and T. Ziemke, pp. 105–110, Berlin, Springer Verlag. Perspectives in Neural Computing.
Under the assumption of asymptotically unbiased estimators it is shown that there exists a nontrivial choice of the insensitivity parameter in Vapnik's epsilon–insensitive loss function which scales linearly with the input noise of the training data. This finding is backed by experimental results.