Loss Functions, Complexities, and the Legendre Transformation
The paper introduces a way of re-constructing a loss function from predictive complexity. We show that a loss function and expectations of the corresponding predictive complexity w.r.t. the Bernoulli distribution are related through the Legendre transformation. It is shown that if two loss functions specify the same complexity then they are equivalent in a strong sense.
Unable to display preview. Download preview PDF.
- V. M. Alekseev, V. M. Tikhomirov, and S. V. Fomin. Optimal Control. Plenum, New York, 1987.Google Scholar
- Y. Kalnishkan. General linear relations among different types of predictive complexity. In Proc. 10th International Conference on Algorithmic Learning Theory — ALT’ 99, volume 1720 of Lecture Notes in Artificial Intelligence, pages 323–334. Springer-Verlag, 1999.Google Scholar
- A. Wayne Roberts and Dale E. Varberg. Convex Functions. Academic Press, 1973.Google Scholar
- R. Tyrrell Rockafellar. Convex Analysis. Princeton University Press, 1970.Google Scholar
- V. Vovk. Aggregating strategies. In M. Fulk and J. Case, editors, Proceedings of the 3rd Annual Workshop on Computational Learning Theory, pages 371–383, San Mateo, CA, 1990. Morgan Kaufmann.Google Scholar
- V. Vovk and C. J. H. C. Watkins. Universal portfolio selection. In Proceedings of the 11th Annual Conference on Computational Learning Theory, pages 12–23, 1998.Google Scholar