A B C D E F G H I J K L M N O P Q R S T U V W X Y Z All
Chaturvedi, Soni
- Digits and Special Character Recognition System using ANN and SNN Models
Authors
1 Electronics Engineering, GHRCE, Nagpur, IN
2 Electronics & Communication Engineering, PIET, Nagpur, IN
3 Electronics Department, RCOEM, Nagpur, IN
4 Electronics Department, GHRCE, Nagpur, IN
Source
Digital Image Processing, Vol 6, No 6 (2014), Pagination: 269-275Abstract
The paper depicts pattern recognition of Digits and Special Character using Artificial Neural Network’s model Feed Forward Neural Network using Back Propagation Algorithm and Spiking Neural Network’s model Izhikevich Neuron model. Artificial Neural network is the second generation model and Spiking Neural Network is third generation model of Neural Networks. In this paper we have focused on recognizing the patterns and comparing the results of the two models Feed forward Neural Network and Izhikevich model by following the main steps of pattern recognition which are, Scanned handwritten Images, Pre-Processing of image, Feature Extraction of image, Training the network, Recognition/Classification of images.
Feed Forward Neural Network consists of three layers and this network is trained by Back propagation Algorithm. On the other hand, we have Izhikevich model which is well known for producing all known firing rates pattern. It is a combination of Hodgkin-Huxley model and computationally efficient Integrate-and Fire neuron model. Izhikevich model will recognize the input pattern of the image and generate the spikes. Using the above two models simulation results are obtained for recognition of digits and special characters and then comparing both models based on recognition rate, reliability, simulation time and number of neurons. We show that the Izhikevich model of spiking neural type provides improved performance.
Keywords
Artificial Neural Network (ANN), Feed Forward Neural Network (FFNN), Izhikevich Neuron Model, Spiking Neural Network (SNN).
- Comparison of LIF and Izhikevich Spiking Neural Models for Recognition of Uppercase and Lowercase English Characters
Authors
1 Electronics Engineering, GHRCE, Nagpur, IN
2 Electronics & Communication Engineering, PIET, Nagpur, IN
3 Electronics Department, RCOEM, Nagpur, IN
4 Electronics Department, GHRCE, Nagpur, IN
Source
Digital Image Processing, Vol 6, No 6 (2014), Pagination: 276-281Abstract
This paper depicts pattern classification of uppercase and lower case English character, using Leaky integrated and fire neuron model and Izhikevich neuron model of spiking neural network. Spiking neural network is one of the best artificial neural networks, which are widely used in the field of neuron science. In this paper we focused on spiking mechanism of both models and compare them in terms of accuracy and simulation time.
Leaky integrate and fire neuron model which is one of the simple and efficient model of spiking neural network that analyze and simulate efficiently. On the other hand Izhikevich model is one of the powerful models which can simulate thousands of neurons in real time. Using these two models simulation results are obtained for recognition of uppercase and lower case English characters. Finally we report on simulation results of both models and discuss their performance, in terms of recognition rate and speed.