Erratum: Sgdqn is less careful than expected

Antoine Bordes (1) , Léon Bottou (2) , Patrick Gallinari (1) , J. Chang (3) , S.A. Smith (3)

1 LIP6 - Laboratoire d'Informatique de Paris 6
2 Microsoft Research [Redmond]
3 Facebook

The SGD-QN algorithm described in Bordes et al. (2009) contains a subtle flaw that prevents it from reaching its design goals. Yet the flawed SGD-QN algorithm has worked well enough to be a winner of the first Pascal Large Scale Learning Challenge (Sonnenburg et al., 2008). This document clarifies the situation, proposes a corrected algorithm, and evaluates its performance.

Domaines: Apprentissage [cs.LG]

Antoine Bordes, Léon Bottou, Patrick Gallinari, J. Chang, S.A. Smith. Erratum: Sgdqn is less careful than expected. Journal of Machine Learning Research, 2010, 11, pp.2229--2240. ⟨hal-00750268⟩