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

Learning persistent dynamics with neural networks.

H Waelbroeck1, H D. Navone, H A. Ceccatto

  • 1Instituto de Ciencias Nucleares, UNAM, Mexico

Neural Networks : the Official Journal of the International Neural Network Society
|March 29, 2003
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

Symmetry breaking and adaptation: evidence from a 'toy model' of a virus.

Bio Systems·1999
Same author

Codon bias and mutability in HIV sequences.

Journal of molecular evolution·1999
Same author

Self-adaptation in evolving systems.

Artificial life·1998
Same author

Emergence of algorithmic language in genetic systems.

Bio Systems·1998
Same journal

Dynamic analysis and reliable mechanical optimization application of ring HNN effected with a memristive neuron.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

DAFF-Net: A detection and search method for small-scale low surface brightness galaxies.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Quasi-synchronization for complex networks with hybrid pinning intermittent control.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Physics-encoded convolutional neural operators for parametric PDEs: A convergence-guaranteed framework via pre-computed kernel fields.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles

Neural networks struggle with persistent time series data, sometimes getting stuck in error minima. Training can be abrupt for chaotic data but smooth for noisy data, with decorrelation methods offering potential solutions.

Area of Science:

  • Machine Learning
  • Time Series Analysis
  • Dynamical Systems

Background:

  • Real-world time series frequently exhibit persistence.
  • Neural networks are increasingly used for time series analysis.
  • Understanding learning dynamics in neural networks is crucial.

Purpose of the Study:

  • Investigate neural network learning of persistent time series dynamics.
  • Identify challenges and potential solutions for training on persistent data.
  • Evaluate the effectiveness of decorrelation methods.

Main Methods:

  • Utilized neural networks to model persistent time series.
  • Analyzed training behavior for chaotic and noisy dynamics.
  • Assessed the impact of two common decorrelation techniques.

Related Experiment Videos

Main Results:

  • Neural networks can become trapped in trivial error minima due to long-term autocorrelation in chaotic time series.
  • The transition to a trained state can be abrupt for chaotic data.
  • Training on noisy dynamics results in a smooth learning process.
  • Decorrelation methods were examined for their efficacy in mitigating persistence issues.

Conclusions:

  • Persistence in time series presents unique challenges for neural network training.
  • The nature of the dynamics (chaotic vs. noisy) significantly impacts the learning process.
  • Further research into decorrelation methods is warranted for effective persistent time series modeling.