Title :
A performance evaluation of probabilistic vs. deterministic spiking neural network
Author :
Sedghi, Maryam ; Ahmadi, Amin ; Eskandari, Elahe ; Heydari, Ramiyar
Author_Institution :
Electr. Eng. Dept., Razi Univ., Kermanshah, Iran
Abstract :
This paper aims to present a comparison between probabilistic and deterministic spiking neural network for a back Propagation classification algorithm. To have a fair comparison, neuron models and structures are considered identical in both of the networks. The networks are trained and tested with the Iris database. According to the simulation results, the probabilistic network converges faster than the deterministic one, where it is also more sensitive to the input variations. The simulation results show a precision of 90% 88% for the probabilistic and the deterministic networks correspondingly, which is in consistence with the similar results for linear neural networks.
Keywords :
backpropagation; neural nets; pattern classification; probability; back propagation classification algorithm; deterministic spiking neural network; linear neural networks; neuron models; neuron structures; performance evaluation; probabilistic spiking neural network; Biological neural networks; Biological system modeling; Encoding; Mathematical model; Neurons; Probabilistic logic; Training; Back Propagation Algorithm; Probabilistic Spiking Neural Network (PSNN); Spiking Neural Network (SNN);
Conference_Titel :
Electrical Engineering (ICEE), 2014 22nd Iranian Conference on
Conference_Location :
Tehran
DOI :
10.1109/IranianCEE.2014.6999547
