Skip to Main content Skip to Navigation
Conference papers

SeaRNN: Training RNNs with Global-Local Losses

Rémi Leblond 1 Jean-Baptiste Alayrac 2 Anton Osokin 1 Simon Lacoste-Julien 2
1 SIERRA - Statistical Machine Learning and Parsimony
DI-ENS - Département d'informatique de l'École normale supérieure, CNRS - Centre National de la Recherche Scientifique, Inria de Paris
Abstract : We propose SEARNN, a novel training algorithm for recurrent neural networks (RNNs) inspired by the "learning to search" (L2S) approach to structured prediction. RNNs have been widely successful in structured prediction applications such as machine translation or parsing, and are commonly trained using maximum likelihood estimation (MLE). Unfortunately, this training loss is not always an appropriate surrogate for the test error: by only maximizing the ground truth probability, it fails to exploit the wealth of information offered by structured losses. Further, it introduces discrepancies between training and predicting (such as exposure bias) that may hurt test performance. Instead, SEARNN leverages test-alike search space exploration to introduce global-local losses that are closer to the test error. We first demonstrate improved performance over MLE on two different tasks: OCR and spelling correction. Then, we propose a subsampling strategy to enable SEARNN to scale to large vocabulary sizes. This allows us to validate the benefits of our approach on a machine translation task.
Complete list of metadata

Cited literature [32 references]  Display  Hide  Download

https://hal.inria.fr/hal-01950555
Contributor : Rémi Leblond <>
Submitted on : Monday, December 10, 2018 - 10:47:59 PM
Last modification on : Tuesday, May 4, 2021 - 2:06:02 PM
Long-term archiving on: : Monday, March 11, 2019 - 5:26:28 PM

File

15c49b74558c239a4dbfa438d3abe7...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01950555, version 1

Collections

Citation

Rémi Leblond, Jean-Baptiste Alayrac, Anton Osokin, Simon Lacoste-Julien. SeaRNN: Training RNNs with Global-Local Losses. ICLR 2018 : 6th International Conference on Learning Representations, Apr 2018, Vancouver, Canada. ⟨hal-01950555⟩

Share

Metrics

Record views

101

Files downloads

260