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

Backpropagation uses prior information efficiently.

E Barnard1, E C Botha

  • 1Dept. of Comput. Sci. and Eng., Oregon Graduate Inst., Portland, OR.

IEEE Transactions on Neural Networks
|January 1, 1993
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

Consent and Chaperone Practices in Transvaginal Ultrasound: Exploring Sonographer Knowledge, Gender, and Ethical Challenges.

Irish medical journal·2026
Same author

Clinical, tactical and strategic benefits of a UK Spray Dried Plasma production capability.

BMJ military health·2025
Same author

Oral tranexamic acid as a preferred administration route for severe trauma in the extreme cold weather environment.

BMJ military health·2025
Same author

'Golden day' is a myth: rethinking medical timelines and risk in large scale combat operations.

BMJ military health·2024
Same author

Prehospital emergency care in a humanitarian environment: an overview of the ethical considerations.

BMJ military health·2023
Same author

Red cell haemolysis secondary to intraosseous (IO) blood transfusion in adult patients with major trauma: a systematic review.

BMJ military health·2023
Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013
Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011
Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011
Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011
Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011
Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011
See all related articles

Neural network classifiers use prior information suboptimally, but this has minor impacts on performance. Excessive network parameters can worsen generalization due to this suboptimality.

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Computer Science

Background:

  • Neural network classifiers are widely used in various applications.
  • Understanding how neural networks utilize a priori information is crucial for optimizing their performance.
  • Prior information, such as known data distributions, can potentially improve classifier accuracy.

Purpose of the Study:

  • To investigate the capability of neural network classifiers to incorporate a priori information.
  • To evaluate the impact of suboptimal use of a priori information on classifier performance and generalization.
  • To determine the relationship between network complexity, a priori information utilization, and generalization errors.

Main Methods:

  • Training backpropagation classifiers using datasets with known distributions and varying a priori probabilities.

Related Experiment Videos

  • Evaluating classifier performance on independent test sets to assess generalization.
  • Analyzing the extent to which a priori information is optimally employed by the backpropagation algorithm.

    • Main Results:

    • Backpropagation classifiers were found to utilize a priori information in a slightly suboptimal manner.
    • This suboptimality did not lead to significant degradation in classifier performance.
    • Inferior generalization observed with overly complex networks was partially attributed to this suboptimal use of a priori information.

    Conclusions:

    • Neural networks exhibit a slight suboptimality in leveraging a priori information, with minimal adverse effects on performance.
    • Network complexity plays a role in generalization, partly due to how a priori information is processed.
    • Further research could explore methods to improve the optimal integration of a priori knowledge in neural networks.