TY - JOUR
T1 - An efficient deep neural network accelerator using controlled ferroelectric domain dynamics
AU - Majumdar, Sayani
N1 - Funding Information:
The author thanks the Academy of Finland (Grant Nos. 345068 and 350667) for financial support. Dr Binbin Chen is acknowledged for performing the KPFM and PFM measurements of the P(VDF-TrFE) films. The work used experimental facilities of Micronova national research infrastructure for micro- and nanotechnology.
PY - 2022/11/28
Y1 - 2022/11/28
N2 - The current work reports an efficient deep neural network (DNN) accelerator, where analog synaptic weight elements are controlled by ferroelectric (FE) domain dynamics. An integrated device-to-algorithm framework for benchmarking novel synaptic devices is used. In poly(vinylidene fluoride-trifluoroethylene)-based ferroelectric tunnel junctions (FTJs), analog conductance states are measured using a custom pulsing protocol, and associated control circuits and array architectures for DNN training are simulated. Our results show that precise control of polarization switching dynamics in multi-domain polycrystalline FE thin films can produce considerable weight-update linearity in metal–ferroelectric–semiconductor (MFS) tunnel junctions. Ultrafast switching and low junction currents in these devices offer extremely energy-efficient operation. Via an integrated platform of hardware development, characterization and modeling, we predict the available conductance range, where linearity is expected under identical potentiating and depressing pulses for efficient DNN training and inference tasks. As an example, an analog crossbar-based DNN accelerator with MFS junctions as synaptic weight elements showed >93% training accuracy on a large MNIST handwritten digit dataset while, for cropped images, >95% accuracy is achieved. One observed challenge is the rather limited dynamic conductance range while operating under identical potentiating and depressing pulses below 1 V. Investigation is underway to improve the FTJ dynamic conductance range, maintaining the weight-update linearity under an identical pulse scheme.
AB - The current work reports an efficient deep neural network (DNN) accelerator, where analog synaptic weight elements are controlled by ferroelectric (FE) domain dynamics. An integrated device-to-algorithm framework for benchmarking novel synaptic devices is used. In poly(vinylidene fluoride-trifluoroethylene)-based ferroelectric tunnel junctions (FTJs), analog conductance states are measured using a custom pulsing protocol, and associated control circuits and array architectures for DNN training are simulated. Our results show that precise control of polarization switching dynamics in multi-domain polycrystalline FE thin films can produce considerable weight-update linearity in metal–ferroelectric–semiconductor (MFS) tunnel junctions. Ultrafast switching and low junction currents in these devices offer extremely energy-efficient operation. Via an integrated platform of hardware development, characterization and modeling, we predict the available conductance range, where linearity is expected under identical potentiating and depressing pulses for efficient DNN training and inference tasks. As an example, an analog crossbar-based DNN accelerator with MFS junctions as synaptic weight elements showed >93% training accuracy on a large MNIST handwritten digit dataset while, for cropped images, >95% accuracy is achieved. One observed challenge is the rather limited dynamic conductance range while operating under identical potentiating and depressing pulses below 1 V. Investigation is underway to improve the FTJ dynamic conductance range, maintaining the weight-update linearity under an identical pulse scheme.
KW - Neuromorphic computing
KW - deep neural networks
KW - ferroelectric tunnel junctions
KW - nonvolatile memory
KW - neuromorphic engineering
KW - deep neural network accelerator
KW - ferroelectric tunnel junction
KW - in-memory computing
KW - ferroelectric domain dynamics
KW - neuromorphic computing
UR - http://www.scopus.com/inward/record.url?scp=85160947145&partnerID=8YFLogxK
U2 - 10.1088/2634-4386/ac974d
DO - 10.1088/2634-4386/ac974d
M3 - Article
SN - 2634-4386
VL - 2
JO - Neuromorphic Computing and Engineering
JF - Neuromorphic Computing and Engineering
IS - 4
M1 - 041001
ER -