Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Temporal constraints on visual learning: a computational model.

J V Stone1, N Harper

  • 1Department of Psychology, University of Sheffield, UK. j.v.stone@shef.ac.uk

Perception
|March 1, 2000
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Localized autosomal recessive epidermolysis bullosa simplex arising from a novel homozygous frameshift mutation in DST (BPAG1).

Clinical and experimental dermatology·2021
Same author

Do perioperative antibiotics reduce the risk of surgical-site infections following excision of ulcerated skin cancers? A Critically Appraised Topic.

The British journal of dermatology·2017
Same author

Specialist perioperative allergy clinic services in the UK 2016: Results from the Royal College of Anaesthetists Sixth National Audit Project.

Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology·2017
Same author

The 'golden goddess' of spices: a potentially effective topical therapy for Bowen disease?

Clinical and experimental dermatology·2017
Same author

Chlorhexidine allergy in four specialist allergy centres in the United Kingdom, 2009-13: clinical features and diagnostic tests.

Clinical and experimental immunology·2017
Same author

Survey of dermatologists demonstrates widely varying approaches to perioperative antibiotic use: time for a randomized trial?

The British journal of dermatology·2016
Same journal

Benchmarking spatial discrimination thresholds of two-frame motion defined forms compared to luminance and stereoscopic defined forms.

Perception·2026
Same journal

The effect of face masks on the perception of trustworthiness and competence in individuals with autistic traits.

Perception·2026
Same journal

The importance of external features for categorizing ethnicity: can Koreans identify Korean, Japanese, and Chinese faces?

Perception·2026
Same journal

Interoception, alexithymia, and motor congruency: Psychological drivers of body ownership in virtual reality.

Perception·2026
Same journal

The frustration of a small <i>n</i>.

Perception·2026
Same journal

Why do we have two eyes.

Perception·2026
See all related articles

This study introduces temporal smoothness as a constraint for artificial neural networks, enabling them to learn underlying invariances from perceptual data. This method helps networks learn salient properties like stereo disparity more effectively.

Area of Science:

  • Artificial Intelligence
  • Computer Vision
  • Machine Learning

Background:

  • Artificial neural networks often struggle to learn invariances from continuous perceptual data.
  • Superficial image properties encoded by network units can vary rapidly, hindering learning of stable features.

Purpose of the Study:

  • To develop a learning methodology that constrains artificial neural networks to learn underlying invariances from perceptual stimuli.
  • To investigate the effectiveness of temporal smoothness as a learning constraint.

Main Methods:

  • Implemented a temporal-smoothness constraint in a backpropagation network.
  • Formalized temporal smoothness using regularization theory by modifying the network's cost function.
  • Trained the network on random-dot stereograms to learn stereo disparity.

Related Experiment Videos

Main Results:

  • The temporal-smoothness constraint successfully enabled the network to learn stereo disparity.
  • Temporal smoothness demonstrated effectiveness comparable to other generalization techniques like early stopping and weight decay.
  • The theoretical basis of temporal smoothing is strongly linked to physical world characteristics.

Conclusions:

  • Temporal smoothness is a viable and theoretically grounded method for improving generalization in artificial neural networks.
  • This approach leverages fundamental properties of the physical world for more robust learning of perceptual invariances.