Margin conditions for vector quantization

Clément Levrard (1, 2, 3)

1 UP11 - Université Paris-Sud - Paris 11
2 UPMC - Université Pierre et Marie Curie - Paris 6
3 SELECT - Model selection in statistical learning

In this report, oracle inequalities on the excess risk of the empirical risk minimizer in the quantization framework are derived. These inequalities are based on conditions which may be thought of as margin type conditions, such as one derived in the statistical learning framework. Furthermore, these inequalities derive from innovative chaining techniques and its use for Dudley's entropy integral.

Mots clés: quantization; fast rates; margin conditions

Domaines: Statistiques [math.ST]; Théorie [stat.TH]

Clément Levrard. Margin conditions for vector quantization. 2014. ⟨hal-00877093v2⟩