Abstract

Neuromorphic systems (hardware neural networks) derive inspiration from biological neural systems and are expected to be a computing breakthrough beyond conventional von Neumann architecture. Interestingly, in neuromorphic systems, the processing and storing of information can be performed simultaneously by modulating the connection strength of a synaptic device (i.e., synaptic weight). Previously investigated synaptic devices can emulate the functionality of biological synapses successfully by utilizing various nano-electronic phenomena; however, the impact of intrinsic synaptic device variability on the system performance has not yet been studied. Here, we perform a device-to-system level simulation of different synaptic device variation parameters in a designed neuromorphic system that has the potential for unsupervised learning and pattern recognition. The effects of variations in parameters such as the weight modulation nonlinearity (NL), the minimum-maximum weight (Gmin and Gmax), and the weight update margin (ΔG) on the pattern recognition accuracy are analyzed quantitatively. These simulation results can provide guidelines for the continued design and optimization of a synaptic device for realizing a functional large-scale neuromorphic computing system.

Highlights

  • To date, various nano-electronic devices have successfully reproduced a specific learning rule of biological synapses through their internal analog conductance states that can be modulated intentionally with an applied pulse’s timing or level[7,8,9,10,11,12,13,14,15]

  • A device-to-system level simulation presents quantitative results in terms of the neuromorphic system performance depending on different device variation parameters, where the discussed system has the potential for unsupervised online learning and consequent image classification for MNIST handwritten digits

  • Note that our analysis is only limited to the neuromorphic system that are capable of online learning

Read more

Summary

Accuracy in a Hardware Neural

Neuromorphic systems (hardware neural networks) derive inspiration from biological neural systems and are expected to be a computing breakthrough beyond conventional von Neumann architecture. The effects of variations in parameters such as the weight modulation nonlinearity (NL), the minimum-maximum weight (Gmin and Gmax), and the weight update margin (ΔG) on the pattern recognition accuracy are analyzed quantitatively. These simulation results can provide guidelines for the continued design and optimization of a synaptic device for realizing a functional large-scale neuromorphic computing system. The effect of the synaptic device variations in the weight modulation nonlinearity (NL), the minimum-maximum weight (Gmin and Gmax), and the weight update margin (ΔG) on the pattern recognition accuracy is analyzed These simulation results provide a design guideline for the specifications of the synaptic device needed for more reliable system operation. The comprehensive study for the effects of device variation to the system performance with online-learning has been missing[29,30,31,32], our study aims the quantitative analysis to understand how the pattern recognition accuracy of the existing STDP-based online-learning system affected by device variation parameters

Results and Discussion
Methods
Additional Information
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call