Multi-neuron connection using multi-terminal floating-gate memristor for unsupervised learning.
Ui Yeon Won, Quoc An Vu, Sung Bum Park, Mi Hyang Park, Van Dam Do, Hyun Jun Park, Heejun Yang, Young Hee Lee, Woo Jong Yu
Author Information
Ui Yeon Won: Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, 16419, South Korea.
Quoc An Vu: IBS Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon, 16419, South Korea.
Sung Bum Park: Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, 16419, South Korea.
Mi Hyang Park: Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, 16419, South Korea.
Van Dam Do: Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, 16419, South Korea.
Hyun Jun Park: Display R&D Group, Mobile Communication Business, Samsung Electronics, Suwon, 16677, South Korea.
Heejun Yang: Department of Physics, Korea Advanced Institute of Science and Technology, Daejeon, 34141, South Korea. ORCID
Young Hee Lee: IBS Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon, 16419, South Korea. leeyoung@skku.edu. ORCID
Woo Jong Yu: Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, 16419, South Korea. micco21@skku.edu. ORCID
Multi-terminal memristor and memtransistor (MT-MEMs) has successfully performed complex functions of heterosynaptic plasticity in synapse. However, theses MT-MEMs lack the ability to emulate membrane potential of neuron in multiple neuronal connections. Here, we demonstrate multi-neuron connection using a multi-terminal floating-gate memristor (MT-FGMEM). The variable Fermi level (E) in graphene allows charging and discharging of MT-FGMEM using horizontally distant multiple electrodes. Our MT-FGMEM demonstrates high on/off ratio over 10 at 1000 s retention about ~10,000 times higher than other MT-MEMs. The linear behavior between current (I) and floating gate potential (V) in triode region of MT-FGMEM allows for accurate spike integration at the neuron membrane. The MT-FGMEM fully mimics the temporal and spatial summation of multi-neuron connections based on leaky-integrate-and-fire (LIF) functionality. Our artificial neuron (150 pJ) significantly reduces the energy consumption by 100,000 times compared to conventional neurons based on silicon integrated circuits (11.7 μJ). By integrating neurons and synapses using MT-FGMEMs, a spiking neurosynaptic training and classification of directional lines functioned in visual area one (V1) is successfully emulated based on neuron's LIF and synapse's spike-timing-dependent plasticity (STDP) functions. Simulation of unsupervised learning based on our artificial neuron and synapse achieves a learning accuracy of 83.08% on the unlabeled MNIST handwritten dataset.
References
Science. 2019 May 10;364(6440):570-574
[PMID: 31023890]
