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

The simultaneous recurrent neural network for addressing the scaling problem in static optimization.

G Serpen1, A Patwardhan, J Geib

  • 1Electrical Engineering and Computer Science Department of the University of Toledo, Toledo, OH 43606, USA. gursel.serpen@ieee.org

International Journal of Neural Systems
|December 26, 2001
PubMed
Summary

A new trainable neural network, Simultaneous Recurrent Neural network (SRNN), effectively solves static optimization scaling problems. This learning-based approach consistently finds high-quality solutions for complex problems like the Traveling Salesman Problem.

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

Teaching molecular genetics using <i>Paramecium</i> and RNA interference: research-based learning and project ownership.

Journal of microbiology & biology education·2025
Same author

Safety, Efficacy and Clinical Applications of Focused Ultrasound-Mediated Blood Brain Barrier Opening in Alzheimer's Disease: A Systematic Review.

The journal of prevention of Alzheimer's disease·2024
Same author

A Biomechanical Analysis of Tibial Fixation Methods in Hamstring-Graft Anterior Cruciate Ligament Reconstruction.

The Iowa orthopaedic journal·2019
Same author

Biologic therapies for refractory juvenile dermatomyositis: five years of experience of the Childhood Arthritis and Rheumatology Research Alliance in North America.

Pediatric rheumatology online journal·2017
Same author

Databases and Archiving for CryoEM.

Methods in enzymology·2016
Same author

Exome sequencing in 32 patients with anophthalmia/microphthalmia and developmental eye defects.

Clinical genetics·2014

Area of Science:

  • Artificial Intelligence
  • Computational Optimization
  • Machine Learning

Background:

  • Static optimization problems present significant scaling challenges for traditional neural network algorithms.
  • Existing recurrent neural network algorithms often lack the trainability required to adapt to complex optimization landscapes.

Purpose of the Study:

  • To introduce a trainable recurrent neural network, the Simultaneous Recurrent Neural network (SRNN), designed to overcome the scaling limitations in static optimization.
  • To demonstrate the SRNN's capability in finding optimal solutions for computationally intensive problems.

Main Methods:

  • The SRNN utilizes recurrent backpropagation for training, enabling its relaxation-based dynamics to converge to locally optimal solutions.
  • The network's performance was evaluated on the NP-hard Traveling Salesman Problem (TSP) across various scales.

Related Experiment Videos

Main Results:

  • The SRNN demonstrated consistent performance in locating high-quality solutions for the Traveling Salesman Problem, even for instances with up to 600 cities.
  • The algorithm exhibits robust scalability with respect to solution quality as problem size increases.
  • Increased computational cost was observed for larger problem instances, a trade-off for enhanced solution quality.

Conclusions:

  • The Simultaneous Recurrent Neural network offers a trainable and scalable solution for static optimization problems.
  • The SRNN's learning capability allows it to effectively address challenges that limit conventional neural network approaches.
  • This research highlights the potential of trainable recurrent networks for tackling complex optimization tasks, including the Traveling Salesman Problem.