Information Processing in Neuron with Exponential Distributed Delay
Journal Title: International Journal of Machine Learning and Networked Collaborative Engineering - Year 2018, Vol 2, Issue 2
Abstract
Artificial intelligence (AI) has been become the primary need in nearly all sectors namely engineering, services, banking, finance, defense, space etc [3], [33]. Artificial intelligence in these sectors can be implemented in two ways: (i) hardware level implementation (ii) software level implementation. Both kinds of AI implementation require neuron models which mimic the minimal set of real neuron functionality. To this end, Leaky Integrate-and-Fire (LIF) model is performing as the backbone for both kinds of AI implementation. At hardware level implementation, it’s a variant, called as neuristors, is used at chip level implementation, whereas a number of variants LIF model are used to implement AI at software level. In this work, the extended LIF model in distributed delay kernel regime is analyzed. The impact of exponentially distributed delay (EDD) memory kernel on spiking activity and steady state membrane potential distribution (SVD) of LIF neuron is investigated. Fokker-Planck equation associated with the considered model is solved to investigate SVD of the neuron in sub-threshold regime, which results Gaussian distribution. In order to study the information processing, spiking activity of the model is investigated, which is further extended to neuronal rate-code scheme. These finding have been compared with simple LIF model with stochastic input. It is evident that steady state membrane potential distribution of the LIF neuron is invariant due to the presence of EDD. Such kinds of neuron models are useful to implement artificial neural networks. To this end, the proposed model can used to implement recurrent neural networks (RNN) with comparatively more accuracy. Similarly, this model can also be investigated in term of chip level implementation of AI.
Authors and Affiliations
Saket Kumar Choudhary
Keywords
Related Articles
- EP ID EP371009
- DOI 10.30991/IJMLNCE.2018v02i02.003
- Views 108
- Downloads 0
Steganalysis for Reversible Data Hiding Based on Neural Networks and Convolutional Neural Networks
Lossless data hiding techniques is a technique that is very interesting. In which there is a large amount of reversible information hidden technologies. This technique makes it possible to restore the original image afte...
An Innovative Approach for Quick Shopping Using QR-Code for Indian Precinct
In the present scenario life style became too fast and in this rushing life style shopping become hectic for everyone. In the era of technical achievement, there are multiple advanced ways available for shopping. In whic...
An Evaluation of BPMN Solutions
The Business Process Model and Notation (BPMN) is the standard used to represent in a graphical way business processes that take place in every kind of organization and business. This paper analyzes three suites, jBPM, B...
Eight Legs Rimless Wheel Robot Model Driven on Level Ground Using one actuator
It is outstanding that a rimless wheel (RW) needs actuators to walk on the level ground. There is the primary test hard to discover an appropriate control framework to accomplish a stable RW movement. There is the model...
Emerging Trends on Big Data & Cloud Computing
In this paper we focused on the emerging trends and various approaches for carrying out analytics on clouds for Big Data application. It revolves around four important analytics and Big Data. We also discussed about the...