Status : Verified
| Personal Name | Huang, Anfu |
|---|---|
| Resource Title | Handwritten digit character recognition using variants of spiking neural P systems |
| Date Issued | 20 December 2020 |
| Abstract | Spiking neural P systems(SN P systems, for short), is a class of distributed and parallel neural-like computation models, inspired from the way neurons communicate by means of spikes. In the last decade, some researchers have developed plenty of SNP systems and its variants to deal with various academic and engineering problems. It has been shown that SN P systems has powerful capability and significant potential for solving real-life problems, so this realm has received more and more attentions and interests from science and engineering communities. Using the variants of SN P systems(SN P systems with learning functions) to recognize printed English letters has been explored in Song, T., P an, L., W u, T., Zheng, P., Wong, M.D., and Rodriguez-Paton, A. (2019).Spiking neural p systems with learning functions. IEEE transactions on nanobioscience, 18(2) : 176–190. This systems is composed of the Input module and the Recognize module. In the Input module an SN P sub system accepts spike trains, which are related to images of printed English letters with various levels of flipping. In the Recognize module the Liquid State Machines(LSMs, which are also SN P systems) are adopted to identify different spike trains inputted through the Input module. Every LSM used here is a rounded hierarchical design, just like a stone dropped into liquid producing a series of ripples, expanding to the surrounding. The synapses in the LSMs are binded with variable weights generated by Hebbian learning functions in the training phase. This system has obtained a higher recognition accuracy comparing with four types of back propagation neural networks, probabilistic neural networks and spiking neural networks. In this paper, the open problem handwritten character recognition by SN P systems left in Song, et al,. 2019 will be explored. We will implement such the function using two kinds of methodologies, the Liquid State Machines(LSMs) and the Artificial Neural Networks(ANNs). Instea |
| Degree Course | MS Computer Science |
| Language | English |
| Keyword | spiking neural P systems; artificial neural networks; liquid state machines; handwritten digit character recognition |
| Material Type | Thesis/Dissertation |
Preliminary Pages
172.57 Kb
Category : I - Has patentable or registrable invention of creation.
Access Permission : Limited Access