Consiglio Nazionale delle Ricerche

Tipo di prodottoArticolo in rivista
TitoloLetter perception emerges from unsupervised deep learning and recycling of natural image features
Anno di pubblicazione2017
Autore/iTestolin, Alberto; Stoianov, Ivilin; Zorzi, Marco
Affiliazioni autoriUniversità di Padova, CNR, Università di Padova
Autori CNR e affiliazioni
  • inglese
AbstractThe use of written symbols is a major achievement of human cultural evolution. However, how abstract letter representations might be learned from vision is still an unsolved problem. Here, we present a large-scale computational model of letter recognition based on deep neural networks, which develops a hierarchy of increasingly more complex internal representations in a completely unsupervised way by fitting a probabilistic, generative model to the visual input. In line with the hypothesis that learning written symbols partially recycles pre-existing neuronal circuits for object recognition, earlier processing levels in the model exploit domain-general visual features learned from natural images, while domain-specific features emerge in upstream neurons following exposure to printed letters. We show that these high-level representations can be easily mapped to letter identities even for noise-degraded images, producing accurate simulations of a broad range of empirical findings on letter perception in human observers. Our model shows that by reusing natural visual primitives, learning written symbols only requires limited, domain-specific tuning, supporting the hypothesis that their shape has been culturally selected to match the statistical structure of natural environments.
Lingua abstractinglese
Altro abstract-
Lingua altro abstract-
Pagine da657
Pagine a664
Pagine totali8
RivistaNature human behaviour Online
Attiva dal 2017
Editore: Springer Nature
Lingua: inglese
ISSN: 2397-3374
Titolo chiave: Nature human behaviour Online
Titolo proprio: Nature human behaviour.
Numero volume della rivista1
Fascicolo della rivistaSeptember
Verificato da refereeSì: Internazionale
Stato della pubblicazionePublished version
Indicizzazione (in banche dati controllate)
  • ISI Web of Science (WOS) (Codice:000418854500018)
  • Scopus (Codice:2-s2.0-85039897817)
Parole chiavehuman behavior, learning algorithm, perception, reading, deep learning
Link (URL, URI)
Titolo parallelo-
Scadenza embargo-
Data di accettazione-
Note/Altre informazioni-
Strutture CNR
  • ISTC — Istituto di scienze e tecnologie della cognizione
Moduli/Attività/Sottoprogetti CNR-
Progetti Europei
Letter perception emerges from unsupervised deep learning and recycling of natural image features (documento privato )
Tipo documento: application/pdf