Multi-bit Graphene Memristive Synapses for High Precision Neuromorphic Computing
Open Access
- Author:
- Schranghamer, Thomas
- Graduate Program:
- Engineering Science and Mechanics
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- April 13, 2020
- Committee Members:
- Saptarshi Das, Thesis Advisor/Co-Advisor
Judith Todd Copley, Program Head/Chair
Michael T Lanagan, Committee Member
Mark William Horn, Committee Member - Keywords:
- RRAM
Memristor
Graphene
Neuromorphic Computing
Neural Networks
VMM
Artificial Synapse - Abstract:
- As the growth of CMOS technology continues to stagnate, it is important to look towards novel approaches to existing systems. The aging von Neumann architecture, the backbone for modern computing, can no longer fulfill society’s processing needs. As a solution, researchers are looking towards humanity’s original computing system: the human brain. However, to accurately mimic the neural networks of the brain, novel device technologies, architectures, and materials are needed. Currently, memristive crossbar architectures are evolving as powerful in-memory computing engines for artificial neural networks (ANNs). However, the limited number of non-volatile conductance states offered by state-of-the-art memristors is a serious concern for their hardware implementation since trained weights must be rounded to the nearest conductance states, introducing error which can significantly limit inference accuracy. Moreover, the incapability of precise weight updates can lead to convergence problems and slowdown of on-chip training. In this investigation, we circumvent these challenges by introducing graphene-based multi-level (> 16) and non-volatile memristive synapses with arbitrarily programmable conductance states which demonstrate desirable retention and programming endurance. Finally, it is demonstrated that graphene memristors enable weight assignment based on k-means clustering, which offers greater computing accuracy when compared with uniform weight quantization for vector matrix multiplication, an essential component for any ANN.