Search research articles

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

Improving prototypical parts abstraction for case-based reasoning explanations designed for the kidney stone type recognition.

Artificial intelligence in medicine·2025

Same author

Prediction of physical realizations of the coordinated universal time with gated recurrent unit.

The Review of scientific instruments·2024

Same author

A plug-and-play Hyperspectral Imaging Sensor using low-cost equipment.

HardwareX·2022

Same author

Discovery of possible gene relationships through the application of self-organizing maps to DNA microarray databases.

PloS one·2014

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Oct 2, 2025

Author Spotlight: Advancing 3D Cytoarchitecture Analysis - Rapid Volumetric Reconstruction of the Human Brain

Author Spotlight: Advancing 3D Cytoarchitecture Analysis - Rapid Volumetric Reconstruction of the Human Brain

Published on: January 26, 2024

Deep Learning Architecture Reduction for fMRI Data.

Ruben Alvarez-Gonzalez¹, Andres Mendez-Vazquez¹

¹Department of Computer Science, Cinvestav Guadalajara, Zapopan 45015, Mexico.

|February 25, 2022

Summary

This study introduces a texture amortization map (TAM) to enhance feature extraction in deep convolutional neural networks (CNNs). A novel geometric classification score (GCS) measures layer learnability, optimizing CNN performance and reducing model weights.

Keywords:

CNN computer vision deep learning machine learning transfer learning

More Related Videos

Network Analysis of the Default Mode Network Using Functional Connectivity MRI in Temporal Lobe Epilepsy

Network Analysis of the Default Mode Network Using Functional Connectivity MRI in Temporal Lobe Epilepsy

Published on: August 5, 2014

Modeling the Functional Network for Spatial Navigation in the Human Brain

Modeling the Functional Network for Spatial Navigation in the Human Brain

Published on: October 13, 2023

Related Experiment Videos

Last Updated: Oct 2, 2025

Author Spotlight: Advancing 3D Cytoarchitecture Analysis - Rapid Volumetric Reconstruction of the Human Brain

Author Spotlight: Advancing 3D Cytoarchitecture Analysis - Rapid Volumetric Reconstruction of the Human Brain

Published on: January 26, 2024

Network Analysis of the Default Mode Network Using Functional Connectivity MRI in Temporal Lobe Epilepsy

Network Analysis of the Default Mode Network Using Functional Connectivity MRI in Temporal Lobe Epilepsy

Published on: August 5, 2014

Modeling the Functional Network for Spatial Navigation in the Human Brain

Modeling the Functional Network for Spatial Navigation in the Human Brain

Published on: October 13, 2023

Area of Science:

Computer Science
Artificial Intelligence
Machine Learning

Background:

Deep learning models excel at non-linear classification tasks due to advanced architectures.
Optimizing hyper-parameters and kernel selection in deep convolutional neural networks (CNNs) remains challenging for achieving high accuracy.
Deficiencies in algorithms and data processing can hinder classifier performance in practical applications.

Purpose of the Study:

To improve feature extraction in deep learning models by introducing a texture amortization map (TAM).
To develop a novel geometric classification score (GCS) for evaluating layer learnability and optimizing CNN architectures.
To reduce the complexity and computational load of deep learning models by identifying and optimizing key layers.

Main Methods:

Developed an algorithm to extract filter characteristics using a texture amortization map (TAM), considering neighboring pixel textures.
Introduced a geometric classification score (GCS) to quantify the impact of one class on another within classification problems.
Assumed data transformations in inner layers adhere to Euclidean space principles to evaluate layer contributions.

Main Results:

The texture amortization map (TAM) effectively enhances feature extraction capabilities in CNNs.
The geometric classification score (GCS) provides a quantifiable measure of layer learnability and class interaction.
The proposed methods enable the optimization of pre-trained architectures and better model generalization.

Conclusions:

The texture amortization map (TAM) and geometric classification score (GCS) offer novel approaches to improve deep convolutional neural network (CNN) performance.
Understanding feature extraction through TAM and GCS provides insights for optimizing CNN architectures and reducing model complexity.
These methods facilitate more efficient and accurate classification by refining the feature extraction process in deep learning models.