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

Simple and conditioned adaptive behavior from Kalman filter trained recurrent networks.

Lee A Feldkamp1, Danil V Prokhorov, Timothy M Feldkamp

  • 1Research and Advanced Engineering, Ford Motor Company, Dearborn, MI, USA.

Neural Networks : the Official Journal of the International Neural Network Society
|July 10, 2003
PubMed
Summary

Fixed-weight neural networks trained with Kalman filters can mimic adaptive systems. These networks learn past task dependencies and retain this memory even after interference.

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

Model-Free Dual Heuristic Dynamic Programming.

IEEE transactions on neural networks and learning systems·2015
Same author

GrDHP: a general utility function representation for dual heuristic dynamic programming.

IEEE transactions on neural networks and learning systems·2014
Same author

Toyota Prius HEV neurocontrol and diagnostics.

Neural networks : the official journal of the International Neural Network Society·2008
Same author

Special issue on neural networks for feedback control systems.

IEEE transactions on neural networks·2007
Same author

Training recurrent neurocontrollers for real-time applications.

IEEE transactions on neural networks·2007
Same author

Training winner-take-all simultaneous recurrent neural networks.

IEEE transactions on neural networks·2007

Area of Science:

  • Computational neuroscience
  • Machine learning

Background:

  • Adaptive systems are crucial for dynamic environments.
  • Traditional adaptive systems often require complex, explicit mechanisms.
  • Fixed-weight neural networks typically lack inherent adaptability.

Purpose of the Study:

  • To demonstrate that fixed-weight neural networks can exhibit adaptive behaviors.
  • To explore the use of Kalman filter training for adaptability.
  • To show memory retention of past tasks in neural networks.

Main Methods:

  • Utilizing a fixed-weight neural network architecture.
  • Employing Kalman filter methods for network training.
  • Designing conditioning tasks to establish past dependencies.
  • Introducing interference tasks to test memory robustness.

Related Experiment Videos

Main Results:

  • The trained network demonstrated input-output behavior dependent on past conditioning tasks.
  • Network's learned behavior persisted despite intervening interference tasks.
  • Kalman filter training enabled the fixed-weight network to simulate adaptive responses.

Conclusions:

  • Fixed-weight neural networks, when trained with Kalman filters, can effectively perform tasks typically requiring adaptive systems.
  • This approach offers a potential pathway to create computationally efficient systems with memory and adaptability.
  • The findings suggest novel applications in areas requiring robust, context-dependent information processing.