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

Initial state training procedure improves dynamic recurrent networks with time-dependent weights.

L Leistritz1, M Galicki, H Witte

  • 1Inst. of Med. Stat., Friedrich-Schiller-Univ., Jena.

IEEE Transactions on Neural Networks
|February 6, 2008
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

<i>Meloidogyne marylandi</i> is Involved in, but not the Primary Cause of Creeping Bentgrass Decline of Putting Greens in Southern California.

Journal of nematology·2024
Same author

Update: The molecular spectrum of virus-associated high-grade B-cell non-Hodgkin lymphomas.

Blood reviews·2024
Same author

Anaesthetic efficacy and postinduction hypotension with remimazolam compared with propofol: a multicentre randomised controlled trial.

Anaesthesia·2024
Same author

[Current status of teaching in lung ultrasound : Query of knowledge, utilization, need, and preferred teaching method].

Medizinische Klinik, Intensivmedizin und Notfallmedizin·2017
Same author

Topographical principles and peculiarities of operative access in lumbar disc extrusion.

Der Orthopade·2017
Same author

[Pre-operative documentation of individual in-patient therapy goals : A medical staff questionnaire].

Der Anaesthesist·2016
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

This study introduces a novel optimal control method for recurrent neural networks to learn continuous trajectories. The approach efficiently trains networks with time-varying weights, demonstrating effectiveness on complex classification tasks.

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Control Theory

Background:

  • Recurrent neural networks (RNNs) are powerful tools for processing sequential data.
  • Learning multiple continuous trajectories with time-varying weights in RNNs presents significant challenges.
  • Existing methods often struggle with the complexity of dynamic weight adjustments and initial state determination.

Purpose of the Study:

  • To develop an efficient algorithm for training recurrent neural networks to learn multiple continuous trajectories.
  • To address the challenge of time-varying weights and unknown initial states in recurrent neural network learning.
  • To transform the learning problem into an optimal control framework.

Main Methods:

  • The learning problem is reformulated as an optimal control problem.

Related Experiment Videos

  • Network weights and initial states are treated as control variables.
  • A learning algorithm based on a variational formulation of Pontryagin's maximum principle is proposed and analyzed for convergence.

    • Main Results:

    • A novel learning algorithm is presented for recurrent neural networks with time-varying weights.
    • The convergence of the proposed algorithm is theoretically investigated under specific assumptions.
    • Numerical examples demonstrate the algorithm's efficiency in learning nontrivial two-class problems.

    Conclusions:

    • The proposed optimal control-based approach offers an effective solution for training RNNs on complex trajectory learning tasks.
    • The method demonstrates efficiency and convergence properties, making it suitable for dynamic learning scenarios.
    • This work advances the application of optimal control principles in recurrent neural network training.