Event-driven spiking neural network simulator based on FPGA
Recently, researchers have shown an increased interest in more biologically realistic neural networks. Spiking Neural Network (SNN) is one of the most widely used methodologies of mimic neural networks. It has been extensively used for Brain-Machine Interface (BMI), dynamic vision detection (DVS), i...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Master by Coursework |
| Language: | English |
| Published: |
Nanyang Technological University
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/158358 |
| _version_ | 1811693173938847744 |
|---|---|
| author | Zou, Zhili |
| author2 | Gwee Bah Hwee |
| author_facet | Gwee Bah Hwee Zou, Zhili |
| author_sort | Zou, Zhili |
| collection | NTU |
| description | Recently, researchers have shown an increased interest in more biologically realistic neural networks. Spiking Neural Network (SNN) is one of the most widely used methodologies of mimic neural networks. It has been extensively used for Brain-Machine Interface (BMI), dynamic vision detection (DVS), image pattern recognition. From a biophysical point of view, neuron behaviors (action potentials) result from currents that pass through ion channels in the cell membrane. It is possible to simulate such a mimic network on circuit design by modeling the stimulus-voltage relationship. Compared with previous neuron networks, SNN can model a dynamical network in continuous real-time, significantly reducing its power consumption with the event-driven nature. In addition, more researchers participate in exploring the learning methodologies for SNN. As an unsupervised learning fashion, Spike Timing Dependent Plasticity (STDP) has achieved more than 94% accuracy on handwriting digits (MNIST dataset). Furthermore, researchers have migrated some excellent algorithms designed for conventional ANN, CNN to fit in the SNN environment and achieved higher accuracy, close to 99% in a supervised fashion. It has been solidly proved that SNN has the potential to catch up with other artificial neural networks.
Keywords: Spiking Neuron Network, Machine Learning, Pattern Recognition, FPGA, Event-driven. |
| first_indexed | 2024-10-01T06:47:29Z |
| format | Thesis-Master by Coursework |
| id | ntu-10356/158358 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T06:47:29Z |
| publishDate | 2022 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1583582023-07-04T17:51:15Z Event-driven spiking neural network simulator based on FPGA Zou, Zhili Gwee Bah Hwee School of Electrical and Electronic Engineering ebhgwee@ntu.edu.sg Engineering::Electrical and electronic engineering::Electronic circuits Recently, researchers have shown an increased interest in more biologically realistic neural networks. Spiking Neural Network (SNN) is one of the most widely used methodologies of mimic neural networks. It has been extensively used for Brain-Machine Interface (BMI), dynamic vision detection (DVS), image pattern recognition. From a biophysical point of view, neuron behaviors (action potentials) result from currents that pass through ion channels in the cell membrane. It is possible to simulate such a mimic network on circuit design by modeling the stimulus-voltage relationship. Compared with previous neuron networks, SNN can model a dynamical network in continuous real-time, significantly reducing its power consumption with the event-driven nature. In addition, more researchers participate in exploring the learning methodologies for SNN. As an unsupervised learning fashion, Spike Timing Dependent Plasticity (STDP) has achieved more than 94% accuracy on handwriting digits (MNIST dataset). Furthermore, researchers have migrated some excellent algorithms designed for conventional ANN, CNN to fit in the SNN environment and achieved higher accuracy, close to 99% in a supervised fashion. It has been solidly proved that SNN has the potential to catch up with other artificial neural networks. Keywords: Spiking Neuron Network, Machine Learning, Pattern Recognition, FPGA, Event-driven. Master of Science (Electronics) 2022-05-18T05:36:21Z 2022-05-18T05:36:21Z 2022 Thesis-Master by Coursework Zou, Z. (2022). Event-driven spiking neural network simulator based on FPGA. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/158358 https://hdl.handle.net/10356/158358 en application/pdf Nanyang Technological University |
| spellingShingle | Engineering::Electrical and electronic engineering::Electronic circuits Zou, Zhili Event-driven spiking neural network simulator based on FPGA |
| title | Event-driven spiking neural network simulator based on FPGA |
| title_full | Event-driven spiking neural network simulator based on FPGA |
| title_fullStr | Event-driven spiking neural network simulator based on FPGA |
| title_full_unstemmed | Event-driven spiking neural network simulator based on FPGA |
| title_short | Event-driven spiking neural network simulator based on FPGA |
| title_sort | event driven spiking neural network simulator based on fpga |
| topic | Engineering::Electrical and electronic engineering::Electronic circuits |
| url | https://hdl.handle.net/10356/158358 |
| work_keys_str_mv | AT zouzhili eventdrivenspikingneuralnetworksimulatorbasedonfpga |