Computationally Efficient Low Power Neuron Model for Digital Brain

Asif Mehmood, Muhammad Javed Iqbal, Hassan Dawood, Ping Guo

2018 14th International Conference on Computational Intelligence and Security (CIS), IEEE, pp. 1–5, 2018

DOI: 10.1109/CIS2018.2018.00009

© 2018 IEEE. The published version is available from IEEE Xplore via the DOI.

Abstract

Modern supercomputers are capable of performing elementary operations of the order of petaflops. This power is comparable with the computational power of a human brain. The power supply required by supercomputers is in megawatts while human brain requires few watts to perform the same elementary operations as the modern supercomputers can. Thus the power consumption of supercomputers can be reduced (about 10⁵ times) by considering the neuromorphic structure of a brain and attempt to mimic it on hardware chip to construct the computer that can function in the same manner as the human brain. In this paper we have proposed to develop the model for single brain cell, that can be replicated to built the whole brain model having computational power of supercomputers but the energy supply comparable with human brain.

Keywords

• Neuron model
• Digital brain
• Low power
• Neuromorphic computing

Cite this work

@inproceedings{mehmood2018computationally,
  title     = {Computationally Efficient Low Power Neuron Model for Digital Brain},
  author    = {Mehmood, Asif and Iqbal, Muhammad Javed and Dawood, Hassan and Guo, Ping},
  booktitle = {2018 14th International Conference on Computational Intelligence and Security (CIS)},
  pages     = {1--5},
  year      = {2018},
  publisher = {IEEE},
  doi       = {10.1109/CIS2018.2018.00009}
}