Rethinking LDA: Moment Matching for Discrete ICA

Anastasia Podosinnikova 1, 2, 3 Francis Bach 2, 3, 1 Simon Lacoste-Julien 2, 3, 1
3 SIERRA - Statistical Machine Learning and Parsimony
DI-ENS - Département d'informatique de l'École normale supérieure, ENS Paris - École normale supérieure - Paris, Inria Paris-Rocquencourt, CNRS - Centre National de la Recherche Scientifique : UMR8548
Abstract : We consider moment matching techniques for estimation in latent Dirichlet allocation (LDA). By drawing explicit links between LDA and discrete versions of independent component analysis (ICA), we first derive a new set of cumulant-based tensors, with an improved sample complexity. Moreover, we reuse standard ICA techniques such as joint diagonalization of tensors to improve over existing methods based on the tensor power method. In an extensive set of experiments on both synthetic and real datasets, we show that our new combination of tensors and orthogonal joint diagonalization techniques outperforms existing moment matching methods.
Complete list of metadatas

Cited literature [28 references]  Display  Hide  Download

https://hal.inria.fr/hal-01225271
Contributor : Anastasia Podosinnikova <>
Submitted on : Friday, November 6, 2015 - 12:03:19 PM
Last modification on : Thursday, February 7, 2019 - 3:49:22 PM
Long-term archiving on: Monday, February 8, 2016 - 12:49:45 PM

File

lda.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01225271, version 1
  • ARXIV : 1507.01784

Collections

Citation

Anastasia Podosinnikova, Francis Bach, Simon Lacoste-Julien. Rethinking LDA: Moment Matching for Discrete ICA. NIPS 2015 - Advances in Neural Information Processing Systems 28, Dec 2015, Montreal, Canada. ⟨hal-01225271⟩

Share

Metrics

Record views

584

Files downloads

220