A little labeling goes a long way: Semi-supervised learning in infancy

Alexander LaTourrette, Sandra R Waxman

Research output: Contribution to journalArticle

Abstract

There is considerable evidence that labeling supports infants' object categorization. Yet in daily life, most of the category exemplars that infants encounter will remain unlabeled. Inspired by recent evidence from machine learning, we propose that infants successfully exploit this sparsely labeled input through “semi-supervised learning.” Providing only a few labeled exemplars leads infants to initiate the process of categorization, after which they can integrate all subsequent exemplars, labeled or unlabeled, into their evolving category representations. Using a classic novelty preference task, we introduced 2-year-old infants (n = 96) to a novel object category, varying whether and when its exemplars were labeled. Infants were equally successful whether all exemplars were labeled (fully supervised condition) or only the first two exemplars were labeled (semi-supervised condition), but they failed when no exemplars were labeled (unsupervised condition). Furthermore, the timing of the labeling mattered: when the labeled exemplars were provided at the end, rather than the beginning, of familiarization (reversed semi-supervised condition), infants failed to learn the category. This provides the first evidence of semi-supervised learning in infancy, revealing that infants excel at learning from exactly the kind of input that they typically receive in acquiring real-world categories and their names.

LanguageEnglish (US)
Article numbere12736
JournalDevelopmental Science
Volume22
Issue number1
DOIs
StatePublished - Jan 1 2019

Fingerprint

Supervised Machine Learning
Names
Learning
Machine Learning

Keywords

  • category learning
  • conceptual development
  • language acquisition
  • language and thought
  • semi-supervised learning

ASJC Scopus subject areas

  • Developmental and Educational Psychology
  • Cognitive Neuroscience

Cite this

@article{6fd96ca9a0024ef48d5999acbfc5807a,
title = "A little labeling goes a long way: Semi-supervised learning in infancy",
abstract = "There is considerable evidence that labeling supports infants' object categorization. Yet in daily life, most of the category exemplars that infants encounter will remain unlabeled. Inspired by recent evidence from machine learning, we propose that infants successfully exploit this sparsely labeled input through “semi-supervised learning.” Providing only a few labeled exemplars leads infants to initiate the process of categorization, after which they can integrate all subsequent exemplars, labeled or unlabeled, into their evolving category representations. Using a classic novelty preference task, we introduced 2-year-old infants (n = 96) to a novel object category, varying whether and when its exemplars were labeled. Infants were equally successful whether all exemplars were labeled (fully supervised condition) or only the first two exemplars were labeled (semi-supervised condition), but they failed when no exemplars were labeled (unsupervised condition). Furthermore, the timing of the labeling mattered: when the labeled exemplars were provided at the end, rather than the beginning, of familiarization (reversed semi-supervised condition), infants failed to learn the category. This provides the first evidence of semi-supervised learning in infancy, revealing that infants excel at learning from exactly the kind of input that they typically receive in acquiring real-world categories and their names.",
keywords = "category learning, conceptual development, language acquisition, language and thought, semi-supervised learning",
author = "Alexander LaTourrette and Waxman, {Sandra R}",
year = "2019",
month = "1",
day = "1",
doi = "10.1111/desc.12736",
language = "English (US)",
volume = "22",
journal = "Developmental Science",
issn = "1363-755X",
publisher = "Wiley-Blackwell",
number = "1",

}

A little labeling goes a long way : Semi-supervised learning in infancy. / LaTourrette, Alexander; Waxman, Sandra R.

In: Developmental Science, Vol. 22, No. 1, e12736, 01.01.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A little labeling goes a long way

T2 - Developmental Science

AU - LaTourrette, Alexander

AU - Waxman, Sandra R

PY - 2019/1/1

Y1 - 2019/1/1

N2 - There is considerable evidence that labeling supports infants' object categorization. Yet in daily life, most of the category exemplars that infants encounter will remain unlabeled. Inspired by recent evidence from machine learning, we propose that infants successfully exploit this sparsely labeled input through “semi-supervised learning.” Providing only a few labeled exemplars leads infants to initiate the process of categorization, after which they can integrate all subsequent exemplars, labeled or unlabeled, into their evolving category representations. Using a classic novelty preference task, we introduced 2-year-old infants (n = 96) to a novel object category, varying whether and when its exemplars were labeled. Infants were equally successful whether all exemplars were labeled (fully supervised condition) or only the first two exemplars were labeled (semi-supervised condition), but they failed when no exemplars were labeled (unsupervised condition). Furthermore, the timing of the labeling mattered: when the labeled exemplars were provided at the end, rather than the beginning, of familiarization (reversed semi-supervised condition), infants failed to learn the category. This provides the first evidence of semi-supervised learning in infancy, revealing that infants excel at learning from exactly the kind of input that they typically receive in acquiring real-world categories and their names.

AB - There is considerable evidence that labeling supports infants' object categorization. Yet in daily life, most of the category exemplars that infants encounter will remain unlabeled. Inspired by recent evidence from machine learning, we propose that infants successfully exploit this sparsely labeled input through “semi-supervised learning.” Providing only a few labeled exemplars leads infants to initiate the process of categorization, after which they can integrate all subsequent exemplars, labeled or unlabeled, into their evolving category representations. Using a classic novelty preference task, we introduced 2-year-old infants (n = 96) to a novel object category, varying whether and when its exemplars were labeled. Infants were equally successful whether all exemplars were labeled (fully supervised condition) or only the first two exemplars were labeled (semi-supervised condition), but they failed when no exemplars were labeled (unsupervised condition). Furthermore, the timing of the labeling mattered: when the labeled exemplars were provided at the end, rather than the beginning, of familiarization (reversed semi-supervised condition), infants failed to learn the category. This provides the first evidence of semi-supervised learning in infancy, revealing that infants excel at learning from exactly the kind of input that they typically receive in acquiring real-world categories and their names.

KW - category learning

KW - conceptual development

KW - language acquisition

KW - language and thought

KW - semi-supervised learning

UR - http://www.scopus.com/inward/record.url?scp=85053522843&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85053522843&partnerID=8YFLogxK

U2 - 10.1111/desc.12736

DO - 10.1111/desc.12736

M3 - Article

VL - 22

JO - Developmental Science

JF - Developmental Science

SN - 1363-755X

IS - 1

M1 - e12736

ER -