The Omniglot challenge

a 3-year progress report

Brenden Lake, Ruslan Salakhutdinov, Joshua B. Tenenbaum

Research output: Contribution to journalReview article

Abstract

Three years ago, we released the Omniglot dataset for one-shot learning, along with five challenge tasks and a computational model that addresses these tasks. The model was not meant to be the final word on Omniglot; we hoped that the community would build on our work and develop new approaches. In the time since, we have been pleased to see wide adoption of the dataset. There has been notable progress on one-shot classification, but researchers have adopted new splits and procedures that make the task easier. There has been less progress on the other four tasks. We conclude that recent approaches are still far from human-like concept learning on Omniglot, a challenge that requires performing many tasks with a single model.

Original languageEnglish (US)
Pages (from-to)97-104
Number of pages8
JournalCurrent Opinion in Behavioral Sciences
Volume29
DOIs
StatePublished - Oct 1 2019

Fingerprint

Learning
Research Personnel
Datasets

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Psychiatry and Mental health
  • Behavioral Neuroscience

Cite this

The Omniglot challenge : a 3-year progress report. / Lake, Brenden; Salakhutdinov, Ruslan; Tenenbaum, Joshua B.

In: Current Opinion in Behavioral Sciences, Vol. 29, 01.10.2019, p. 97-104.

Research output: Contribution to journalReview article

Lake, Brenden ; Salakhutdinov, Ruslan ; Tenenbaum, Joshua B. / The Omniglot challenge : a 3-year progress report. In: Current Opinion in Behavioral Sciences. 2019 ; Vol. 29. pp. 97-104.
@article{a141af8fde584af68d949080b116d149,
title = "The Omniglot challenge: a 3-year progress report",
abstract = "Three years ago, we released the Omniglot dataset for one-shot learning, along with five challenge tasks and a computational model that addresses these tasks. The model was not meant to be the final word on Omniglot; we hoped that the community would build on our work and develop new approaches. In the time since, we have been pleased to see wide adoption of the dataset. There has been notable progress on one-shot classification, but researchers have adopted new splits and procedures that make the task easier. There has been less progress on the other four tasks. We conclude that recent approaches are still far from human-like concept learning on Omniglot, a challenge that requires performing many tasks with a single model.",
author = "Brenden Lake and Ruslan Salakhutdinov and Tenenbaum, {Joshua B.}",
year = "2019",
month = "10",
day = "1",
doi = "10.1016/j.cobeha.2019.04.007",
language = "English (US)",
volume = "29",
pages = "97--104",
journal = "Current Opinion in Behavioral Sciences",
issn = "2352-1546",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - The Omniglot challenge

T2 - a 3-year progress report

AU - Lake, Brenden

AU - Salakhutdinov, Ruslan

AU - Tenenbaum, Joshua B.

PY - 2019/10/1

Y1 - 2019/10/1

N2 - Three years ago, we released the Omniglot dataset for one-shot learning, along with five challenge tasks and a computational model that addresses these tasks. The model was not meant to be the final word on Omniglot; we hoped that the community would build on our work and develop new approaches. In the time since, we have been pleased to see wide adoption of the dataset. There has been notable progress on one-shot classification, but researchers have adopted new splits and procedures that make the task easier. There has been less progress on the other four tasks. We conclude that recent approaches are still far from human-like concept learning on Omniglot, a challenge that requires performing many tasks with a single model.

AB - Three years ago, we released the Omniglot dataset for one-shot learning, along with five challenge tasks and a computational model that addresses these tasks. The model was not meant to be the final word on Omniglot; we hoped that the community would build on our work and develop new approaches. In the time since, we have been pleased to see wide adoption of the dataset. There has been notable progress on one-shot classification, but researchers have adopted new splits and procedures that make the task easier. There has been less progress on the other four tasks. We conclude that recent approaches are still far from human-like concept learning on Omniglot, a challenge that requires performing many tasks with a single model.

UR - http://www.scopus.com/inward/record.url?scp=85066494968&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85066494968&partnerID=8YFLogxK

U2 - 10.1016/j.cobeha.2019.04.007

DO - 10.1016/j.cobeha.2019.04.007

M3 - Review article

VL - 29

SP - 97

EP - 104

JO - Current Opinion in Behavioral Sciences

JF - Current Opinion in Behavioral Sciences

SN - 2352-1546

ER -