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

Parity still isn't a generalisation problem.

R I Damper

    The Behavioral and Brain Sciences
    |March 31, 1999
    PubMed
    Summary
    This summary is machine-generated.

    Generalization failures in parity tasks do not specifically implicate back-propagation or learning algorithms. Different algorithms exhibit unique failure patterns, refuting claims that parity generalization is a key issue for evaluating learning methods.

    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

    Customizing kernel functions for SVM-based hyperspectral image classification.

    IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2008
    Same author

    Neural network models of categorical perception.

    Perception & psychophysics·2000
    Same author

    Determining and improving the fault tolerance of multilayer perceptrons in a pattern-recognition application.

    IEEE transactions on neural networks·1993
    Same author

    A computational model of afferent neural activity from the cochlea to the dorsal acoustic stria.

    The Journal of the Acoustical Society of America·1991
    Same author

    Hand-held text-to-speech device for the non-vocal disabled.

    Journal of biomedical engineering·1987
    Same author

    Television and video technology in the education of deaf children.

    British journal of audiology·1986
    Same journal

    Are language models models?

    The Behavioral and brain sciences·2026
    Same journal

    Large language models illuminate the mechanistic underpinnings of the creative aspect of language use (CALU), long regarded as a mystery.

    The Behavioral and brain sciences·2026
    Same journal

    LLMs as a platform for studying constraint interaction: Motivation and challenges.

    The Behavioral and brain sciences·2026
    Same journal

    Beyond the data gap: Children create languages, violate their input statistics, and exhibit critical periods.

    The Behavioral and brain sciences·2026
    Same journal

    Not-so-strange love: Language models and generative linguistic theories are more compatible than they appear.

    The Behavioral and brain sciences·2026
    Same journal

    Rich data drive generalization: Lessons from machine learning for linguistics and cognitive science.

    The Behavioral and brain sciences·2026
    See all related articles

    Area of Science:

    • Machine Learning
    • Artificial Intelligence
    • Computational Neuroscience

    Background:

    • The debate surrounds whether failures in generalizing parity functions are indicative of fundamental issues with specific learning algorithms, such as back-propagation.
    • Clark & Thornton argue that parity generalization is a relevant problem for evaluating learning algorithms, challenging the author's initial claim.

    Discussion:

    • This commentary critically examines the arguments presented by Clark & Thornton regarding the relevance of parity generalization.
    • It asserts that generalization is not a universal problem but rather algorithm-dependent.
    • Back-propagation is not uniquely implicated by parity generalization failures.

    Key Insights:

    • Failures in generalizing parity functions do not provide generalizable insights into the weaknesses of back-propagation or learning in general.

    Related Experiment Videos

  • Different machine learning algorithms exhibit distinct patterns of failure when faced with generalization tasks.
  • The concept of a 'parity generalization task' is considered oxymoronic, suggesting a flawed premise.

    • Outlook:

    • Future research should focus on algorithm-specific failure modes rather than broad claims based on parity tasks.
    • Understanding diverse generalization patterns across algorithms is crucial for advancing machine learning.
    • This work contributes to a nuanced understanding of learning algorithm limitations and evaluation methodologies.