Hallucinating a Cleanly Labeled Augmented Dataset from a Noisy Labeled Dataset Using GANs

Abstract : Noisy labeled learning methods deal with training datasets containing corrupted labels. However, prediction performances of existing methods on small datasets still leave room for improvements. With this objective, in this paper we present a GAN-based method to generate a clean augmented training dataset from a small and noisy labeled dataset. The proposed approach combines noisy labeled learning principles with GAN state-of-the-art techniques. We demonstrate the usefulness of the proposed approach through an empirical study on simple and complex image datasets.
Complete list of metadatas

Cited literature [21 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02054836
Contributor : Florent Chiaroni <>
Submitted on : Saturday, March 2, 2019 - 2:38:39 PM
Last modification on : Saturday, July 27, 2019 - 3:18:18 PM
Long-term archiving on : Friday, May 31, 2019 - 12:56:10 PM

File

NL_GAN.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02054836, version 1

Citation

F Chiaroni, M-C Rahal, N. Hueber, Frédéric Dufaux. Hallucinating a Cleanly Labeled Augmented Dataset from a Noisy Labeled Dataset Using GANs. 26th IEEE International Conference on Image Processing (ICIP), Sep 2019, Taipei, Taiwan. ⟨hal-02054836⟩

Share

Metrics

Record views

231

Files downloads

208