Jove
Visualize
Contact Us

Related Experiment Video

Updated: Apr 6, 2026

Designing and Implementing Nervous System Simulations on LEGO Robots
10:34

Designing and Implementing Nervous System Simulations on LEGO Robots

Published on: May 25, 2013

15.8K

Designing Artificial Neural Networks Using Particle Swarm Optimization Algorithms.

Beatriz A Garro1, Roberto A Vázquez2

  • 1Instituto en Investigaciones en Matemáticas Aplicadas y en Sistemas, Universidad Nacional Autónoma de México, Ciudad Universitaria, 04510 Mexico City, DF, Mexico.

Computational Intelligence and Neuroscience
|July 30, 2015
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving01:29

Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving

407
Mechanistic models play a crucial role in algorithms for numerical problem-solving, particularly in nonlinear mixed effects modeling (NMEM). These models aim to minimize specific objective functions by evaluating various parameter estimates, leading to the development of systematic algorithms. In some cases, linearization techniques approximate the model using linear equations.
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...
407

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Going with the Flow: Sensorimotor Integration Along the Zebrafish GI Tract.

Cells·2025
Same author

Prognostic factors in cancer patients infected with SARS-CoV-2: a Latin American country results.

Therapeutic advances in chronic disease·2021
Same author

Facing High EEG Signals Variability during Classification Using Fractal Dimension and Different Cutoff Frequencies.

Computational intelligence and neuroscience·2019
Same author

Training spiking neural models using artificial bee colony.

Computational intelligence and neuroscience·2015
Same journal

RETRACTION: Real-Time Modulation of Physical Training Intensity Based on Wavelet Recursive Fuzzy Neural Networks.

Computational intelligence and neuroscience·2026
Same journal

RETRACTION: Multidimensional Heterogeneous Network Link Adaptation Based on Mobile Environment.

Computational intelligence and neuroscience·2026
Same journal

RETRACTION: Framework to Segment and Evaluate Multiple Sclerosis Lesion in MRI Slices Using VGG-UNet.

Computational intelligence and neuroscience·2026
Same journal

RETRACTION: Facial Emotion Recognition Using a Novel Fusion of Convolutional Neural Network and Local Binary Pattern in Crime Investigation.

Computational intelligence and neuroscience·2026
Same journal

RETRACTION: Automatic Intelligent System Using Medical of Things for Multiple Sclerosis Detection.

Computational intelligence and neuroscience·2026
Same journal

RETRACTION: Intangible Cultural Heritage Reproduction and Revitalization: Value Feedback, Practice, and Exploration Based on the IPA Model.

Computational intelligence and neuroscience·2026
See all related articles
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

This article introduces an automated method for building artificial neural networks. By using swarm-based algorithms, the system simultaneously optimizes network structure, internal connections, and mathematical functions. This approach aims to improve efficiency and performance compared to traditional manual design techniques.

Area of Science:

  • Computational intelligence and Particle Swarm Optimization research
  • Machine learning architecture design within computer science

Background:

Designing efficient computational models remains a persistent challenge for researchers in the field. Performance often hinges on specific structural choices and training parameters. Prior work has struggled to balance these interconnected variables effectively. No prior work had resolved the difficulty of tuning architectures alongside synaptic weights simultaneously. That uncertainty drove the development of more adaptive, automated systems. Researchers have long sought methods to minimize manual intervention during the configuration phase. Existing approaches frequently rely on static configurations that limit overall model flexibility. This gap motivated the exploration of swarm-based intelligence for complex network optimization tasks.

Purpose Of The Study:

The aim of this study is to present a methodology that automatically designs an artificial neural network. This research addresses the complexity inherent in configuring network architectures and training parameters manually. The authors seek to evolve synaptic weights, connections, and transfer functions simultaneously. This task is difficult because performance depends on the precise interaction of these three components. The researchers propose using swarm-based algorithms to navigate this high-dimensional search space efficiently. They intend to provide a solution that reduces the need for human intervention in model development. By implementing specific fitness functions, the team hopes to prevent overtraining while maintaining high predictive accuracy. This work investigates whether automated swarm intelligence can outperform traditional manual training techniques in nonlinear classification problems.

Keywords:
computational intelligenceevolutionary algorithmsmodel architecturenonlinear classification

Frequently Asked Questions

The researchers propose that the system simultaneously evolves synaptic weights, network architecture, and neuron transfer functions. This integrated approach aims to optimize the entire model structure rather than tuning individual parameters in isolation.

The team utilizes three distinct swarm-based strategies: Basic Particle Swarm Optimization, the Second Generation of Particle Swarm Optimization, and a novel model termed NMPSO. These tools navigate the search space to identify optimal configurations.

The authors implemented eight distinct fitness functions derived from mean square error and classification error metrics. These functions are necessary to evaluate solution quality while actively discouraging overtraining and excessive connection counts.

The researchers utilize mean square error and classification error data to guide the evolutionary process. These metrics serve as the primary feedback loop for assessing how well the network performs on nonlinear classification problems.

Related Experiment Videos

Last Updated: Apr 6, 2026

Designing and Implementing Nervous System Simulations on LEGO Robots
10:34

Designing and Implementing Nervous System Simulations on LEGO Robots

Published on: May 25, 2013

15.8K

Main Methods:

The review approach involves evaluating three distinct swarm-based algorithms for automated model configuration. Researchers systematically test Basic Particle Swarm Optimization, Second Generation of Particle Swarm Optimization, and the NMPSO model. The team defines eight unique fitness functions to assess candidate solutions during the evolutionary process. These functions incorporate mean square error and classification error to guide the search. The design strategy prioritizes the simultaneous refinement of weights, architecture, and transfer functions. Investigators compare their automated results against manual configurations created via Back-Propagation and Levenberg-Marquardt techniques. The study validates these models using various nonlinear pattern classification problems. This comprehensive evaluation ensures that the proposed framework maintains high accuracy across diverse scenarios.

Main Results:

Key findings from the literature indicate that the automated methodology successfully evolves all three principal network components. The swarm-based models demonstrate the ability to identify optimal synaptic weights, connections, and transfer functions concurrently. Researchers report that the eight proposed fitness functions effectively mitigate risks associated with overtraining. The data shows a significant reduction in connection density compared to standard manual design approaches. The proposed NMPSO model exhibits robust performance across various nonlinear pattern classification challenges. Comparisons reveal that these automated designs achieve accuracy levels competitive with traditional Back-Propagation and Levenberg-Marquardt learning algorithms. The results suggest that simultaneous optimization yields more efficient network architectures. These findings highlight the effectiveness of swarm intelligence in streamlining complex computational design tasks.

Conclusions:

The authors propose that their automated methodology successfully evolves multiple network components simultaneously. These swarm-based strategies demonstrate competitive performance against traditional manual training techniques. The researchers suggest that integrating specific error-based fitness functions helps prevent model overtraining. Their findings indicate that reducing connection density improves the overall efficiency of the resulting architectures. The study confirms that swarm-based models handle nonlinear classification tasks with high accuracy. These results imply that automated design processes can outperform standard back-propagation methods in specific scenarios. The authors conclude that their new model provides a robust framework for complex system configuration. This synthesis highlights the potential for swarm intelligence to streamline neural network development cycles.

The study measures performance by comparing the automated designs against networks configured via Back-Propagation and Levenberg-Marquardt algorithms. These manual benchmarks provide a standard for evaluating the accuracy and efficiency of the swarm-based approach.

The authors propose that their automated framework effectively reduces manual design labor. They claim this methodology offers a viable alternative to traditional training algorithms for solving complex, nonlinear pattern classification challenges.