Label-consistent sparse auto-encoders

Thomas Rolland (1) , Adrian Basarab (2, 3) , Thomas Pellegrini (4)

1 INESC-ID - Instituto de Engenharia de Sistemas e Computadores Investigação e Desenvolvimento em Lisboa
2 IRIT-MINDS - CoMputational imagINg anD viSion
3 UT3 - Université Toulouse III - Paul Sabatier
4 IRIT-SAMoVA - Équipe Structuration, Analyse et MOdélisation de documents Vidéo et Audio

Auto-encoders (AE) is a particular type of unsupervised neural networks that aim at providing a compact representation of a signal or an image [1]. Such AEs are useful for data compression but most of the time the representations they provide are not appropriate as is for a downstream classification task. This is due to the fact that they are trained to minimize a reconstruction error and not a classification loss. Classification attempts with AEs have already been proposed such as contractive AEs [2], correspondence AEs [3] and stacked similarity-aware AEs [4], for instance. Inspired by label-consistent K-SVD (LC-KSVD) [5], we propose a novel supervised version of AEs that integrates class information within the encoded representations.

Mots clés: Neural networks; Data compression

Domaines: Base de données [cs.DB]

Thomas Rolland, Adrian Basarab, Thomas Pellegrini. Label-consistent sparse auto-encoders. Workshop on Signal Processing with Adaptative Sparse Structured Representations (SPARS 2019), Jul 2019, Toulouse, France. pp.1-2. ⟨hal-02419439⟩