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 Video

Updated: Jan 1, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.2K

An energy efficient time-mode digit classification neural network implementation.

O C Akgun1, J Mei2

  • 1Section Bioelectronics, Department of Microelectronics, Delft University of Technology, The Netherlands.

Philosophical Transactions. Series A, Mathematical, Physical, and Engineering Sciences
|December 24, 2019
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

[A brief history of the use of Gypsum in orthopedics].

Zhonghua yi shi za zhi (Beijing, China : 1980)·2024
Same author

[Expression and significance of INSM1 and SOX11 in pancreatic neuroendocrine tumor and solid pseudopapillary neoplasm].

Beijing da xue xue bao. Yi xue ban = Journal of Peking University. Health sciences·2023
Same author

[The identification process of the state of fractures by orthopedic surgeons before radiography].

Zhonghua yi shi za zhi (Beijing, China : 1980)·2023
Same author

[Hypoxia promotes differentiation of human induced pluripotent stem cells into embryoid bodies <i>in vitro</i>].

Nan fang yi ke da xue xue bao = Journal of Southern Medical University·2022
Same author

Zinc oxide nanoparticles promotes liver cancer cell apoptosis through inducing autophagy and promoting p53.

European review for medical and pharmacological sciences·2021
Same author

Precision Measurement of the Beam-Normal Single-Spin Asymmetry in Forward-Angle Elastic Electron-Proton Scattering.

Physical review letters·2020
Same journal

Inverse FIP effect plasma in the solar atmosphere: a synthesis of current understanding and new insights from AR 11967.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2026
Same journal

Signs of sulfur fractionation under high magnetic field strength.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2026
Same journal

First ionization potential fractionation of sulfur observed with spectral imaging of the coronal environment.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2026
Same journal

Chromospheric dynamics and turbulence regulate the solar FIP effect.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2026
Same journal

Exploring the link between wave activity in the photospheric velocity driver and the FIP bias in the solar corona.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2026
Same journal

Radiative hydrodynamic simulations of first ionization potential fractionation in solar flares.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2026
See all related articles

This study introduces an ultra-low energy neural network using time-mode signal processing for handwritten digit classification. The novel design achieves 88% accuracy while consuming minimal energy, paving the way for energy-autonomous computing.

Area of Science:

  • Computer Science
  • Electrical Engineering
  • Artificial Intelligence

Background:

  • Traditional neural networks demand significant power, limiting their use in energy-constrained applications.
  • Developing energy-efficient computing paradigms is crucial for the advancement of intelligent systems and edge devices.

Purpose of the Study:

  • To design and implement an ultra-low energy neural network utilizing time-mode signal processing.
  • To demonstrate the feasibility of this approach for handwritten digit classification using the MNIST dataset.

Main Methods:

  • The study employed monostable multivibrator-based multiplying analogue-to-time converters, fixed-width pulse generators, and digital gates for time-mode operation.
  • A single-layer artificial neural network (ANN) with a Softmin-based activation function was designed and implemented in a 0.18 μm CMOS IC process.
Keywords:
classificationenergy efficiencyhandwritten digitneural networktime-modeultra-low energy

Related Experiment Videos

Last Updated: Jan 1, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.2K
  • The system was tested on the MNIST database with quantized neuron weights.

    • Main Results:

    • The time-mode ANN achieved a classification accuracy of 88% for handwritten digits.
    • The system demonstrated ultra-low energy consumption, dissipating only 65.74 pJ per classification.
    • A classification speed of 2.37k classifications per second was recorded at a supply voltage of 0.6 V.

    Conclusions:

    • The designed time-mode neural network offers a viable solution for ultra-low energy intelligent computing.
    • This approach significantly reduces power consumption compared to conventional methods, enabling energy-autonomous intelligent systems.
    • The results highlight the potential of time-mode signal processing in advancing energy-efficient artificial intelligence.