Neuromorphic speech systems using advanced ReRAM-based synapse

Park, S.; Sheri, Ahmad Muqeem; Kim, J.; Noh, Jinwoo; Jang, Junwon; Jeon, Moongu; Lee, B.; Lee, B. R.; Lee, B. H.; Hwang, Hyunsang

doi:10.1109/iedm.2013.6724692

Cited by 152 publications

(123 citation statements)

References 4 publications

Supporting

Mentioning

121

Contrasting

Order By: Relevance

“…In particular, Pr 0.7 Ca 0.3 MnO 3 (PCMO)-based ReRAM can be utilized as a synapse device on the basis of its demonstrated reliable characteristics. 11,12,18 According to previously reported studies, PCMO-based synapse devices (PCMO-synapse) exhibit spike-timingdependent plasticity (STDP) which is a change in the conductance of the PCMO-synapse by temporally modified pre-and post-neuron spikes, as shown in Fig. 1(a).…”

mentioning

confidence: 89%

“…Moreover, under a positive reset bias (V reset ), the resistance changed from the LRS to the HRS (reset process). 11,12 On the basis of the results from the I-V characteristics, the Al/PCMO sample exhibited a larger operation voltage ratio (¼V set /V reset ) and a larger on/off ratio (the ratio between the conductances in the LRS and HRS at the read voltage), as shown in Figs. 3(a)-3(e).…”

mentioning

confidence: 95%

“…[10][11][12][13][15][16][17] Even though each synapse device has its advantages, the synapse device needs a simple two-terminal structure for high density, low-power operation, an analogous conductance change, and reliable characteristics for the implementation of the biological-brain-like S-HNN. 5 In these respects, ReRAM can be a promising candidate.…”

mentioning

confidence: 99%

“…1(a). 11,12 When the pre-neuron spike is applied before the post-neuron spike, the conductance of the PCMO-synapse increases (potentiation). On the contrary, the conductance of the PCMO-synapse decreases when the post-neuron spike is applied before the preneuron spike (depression).…”

mentioning

confidence: 99%

“…In contrast to depression, the oxygen atoms of the induced oxide move back to PCMO during potentiation, and the conductance is changed from low to high when a negative bias is applied to the TE. 11,12,[21][22][23][24] Considering an implementation of the S-HNN including a high-density and power-efficient synapse device, the redox reaction is the main key factor for improving the synaptic characteristics of the PCMO-synapse. In other words, the redox reaction between the reactive TE and the PCMO layer needs to be well-modified to obtain improved synaptic characteristics.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Trade-off between number of conductance states and variability of conductance change in Pr0.7Ca0.3MnO3-based synapse device

Lee

Moon

Park

et al. 2015

Applied Physics Letters

View full text Add to dashboard Cite

The characteristics of Pr0.7Ca0.3MnO3 (PCMO)-based resistive change memory were evaluated as a synapse device for a emerging computing system inspired by a biological brain. The evaluated samples structured with various reactive top electrodes exhibited a dependence on the metal-oxide free energy. More conductance states which can improve the performance of the brain-inspired computing system were achieved in a sample having a low metal-oxide free energy. During the increase and decrease in the conductance, the low metal-oxide free energy also resulted in an asymmetric conductance change leading to degradation in the computational accuracy of the brain-inspired computing system. The results demonstrated a trade-off between the number of conductance states and the variation in the conductance change in the PCMO-based synapse device.

show abstract

mentioning

confidence: 89%

mentioning

confidence: 95%

mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Trade-off between number of conductance states and variability of conductance change in Pr0.7Ca0.3MnO3-based synapse device

Lee

Moon

Park

et al. 2015

Applied Physics Letters

View full text Add to dashboard Cite

show abstract

Theory and Technology of Memristive Devices

Tetzlaff

Ascoli

Wouters

2022

Wiley Encyclopedia of Electrical and Electronics Engineering

View full text Add to dashboard Cite

CMOS scaling is progressively and inevitably approaching atomic boundaries. The industry is thus exploring the potential of novel disruptive nanotechnologies, based on multifunctional materials, which may allow the development of integrated circuits keeping the trend predicted by Moore in 1965 in the years to come, despite further transistor shrinking being no longer possible. Memristors, originally postulated only theoretically out of a brilliant line of thought by Chua in 1971 and identified at the nanoscale almost 40 years later, back in 2008, by a team of engineers supervised by R.S. Williams at Hewlett Packard Labs, represent the key nanotechnology of the future, enabling novel forms of computation, which are closer to the principles underlying the functionalities of the human brain, given the extraordinarily peculiar capability of these devices to sense, store, and process data in the same physical miniaturized volume. Achieving progress in memristor circuit design crucially calls for the development of synergies among researchers with different academic backgrounds. In fact, only combining the technical knowledge, acquired by memristor theoreticians, with the practical know‐how of experimenters will enable to foster progress in memristor circuit and system development in the near future. In line with this necessity, the present article is the fruit of a joint cooperation between two leading memristor research units from Germany, one based at TU Dresden, which has contributed to establish solid theoretical foundations on resistance switching memories and their circuits since 2008, and the other one based at RWTH Aachen University, which has been the leading experimental research unit on nanostructures of this kind over the same time span. This article first presents a comprehensive overview of the theory of memristors, starting off from the early achievements in the pioneering studies of their father Chua, then classifies and analyses in depth the main physical mechanisms underlying the inherently nonlinear behavior of their physical realizations, and, finally, discusses their most promising applications, from the development of new nonvolatile memories based on resistance switching phenomena to the realization of novel neuromorphic circuits, capable to mimic the complex dynamics of biological entities closer than conventional purely CMOS hardware, without leaving aside an excursus on the potential to leverage the unique features of these devices for the circuit implementation of unconventional brain‐inspired time‐ and energy‐efficient sensing and memcomputing strategies, which is a great appeal for the realization of portable light‐weight technical systems for the Internet‐of‐Things industry, especially but not only to the benefit of healthcare, well‐being, and automotive market sectors.

show abstract

Associative Memory for Image Recovery with a High‐Performance Memristor Array

Zhou

Gao

et al. 2019

Adv Funct Materials

View full text Add to dashboard Cite

Associative memory is one of the significant characteristics of the biological brain. However, it has yet to be realized in a large memristor array due to the high requirements on the memristor device. In this work, the multilevel memristor cell is optimized by employing an electro-thermal modulation layer. Memristor devices show both high resistance, cell-to-cell uniformity, and multilevel resistive switching behaviors with good reliability. A Hopfield neural network is experimentally demonstrated on a 1k memristor array that is capable of realizing the associative memory function for emotion image recovery. By using both asynchronous and synchronous refresh schemes, complete emotion images can be recalled from partial information.

show abstract

Neuromorphic speech systems using advanced ReRAM-based synapse

Cited by 152 publications

References 4 publications

Trade-off between number of conductance states and variability of conductance change in Pr0.7Ca0.3MnO3-based synapse device

Trade-off between number of conductance states and variability of conductance change in Pr0.7Ca0.3MnO3-based synapse device

Theory and Technology of Memristive Devices

Associative Memory for Image Recovery with a High‐Performance Memristor Array

Contact Info

Product

Resources

About