Rethinking deep active learning: Using unlabeled data at model training

Abstract : Active learning typically focuses on training a model on few labeled examples alone, while unlabeled ones are only used for acquisition. In this work we depart from this setting by using both labeled and unlabeled data during model training across active learning cycles. We do so by using unsupervised feature learning at the beginning of the active learning pipeline and semi-supervised learning at every active learning cycle, on all available data. The former has not been investigated before in active learning, while the study of latter in the context of deep learning is scarce and recent findings are not conclusive with respect to its benefit. Our idea is orthogonal to acquisition strategies by using more data, much like ensemble methods use more models. By systematically evaluating on a number of popular acquisition strategies and datasets, we find that the use of unla-beled data during model training brings a surprising accuracy improvement in image classification, compared to the differences between acquisition strategies. We thus explore smaller label budgets, even one label per class.
Document type :
Preprints, Working Papers, ...
Complete list of metadatas

Cited literature [45 references]  Display  Hide  Download

https://hal.inria.fr/hal-02372102
Contributor : Yannis Avrithis <>
Submitted on : Wednesday, November 20, 2019 - 11:26:38 AM
Last modification on : Friday, November 22, 2019 - 1:20:15 AM

File

1911.08177.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02372102, version 1
  • ARXIV : 1911.08177

Citation

Oriane Siméoni, Mateusz Budnik, Yannis Avrithis, Guillaume Gravier. Rethinking deep active learning: Using unlabeled data at model training. 2019. ⟨hal-02372102⟩

Share

Metrics

Record views

11

Files downloads

47