Strategies to Improve the Accuracy of Memristor-Based Convolutional Neural Networks
The DOI matches its Crossref record; 40 of the 41 references it cites were found in a registry or an index.
Abstract
In this article, we quantify several nonideal characteristics of memristor synaptic devices, such as the limited conductance states, write nonlinearities, and variations, and comprehensively investigate their effects on the convolutional neural network (CNN) performance. Our result shows that the available conductance states ( ${N}_{\text {state}}$ ), asymmetric write nonlinearities, and cycle-to-cycle (C2C) variation are critical factors to the learning accuracy, while symmetric write nonlinearities and device-to-device variation go trivial. We accordingly propose three strategies to mitigate their impacts on CNN performance: 1) limiting the weight range to improve the utilization of ${N}_{\text {state}}$ ; 2) adopting a new “with-read” update scheme to mitigate the effects of asymmetric write nonlinearities; and 3) employing multiple memristors for each kernel element to alleviate the impact of C2C variation. Our work would provide guidance for the hardware implementation and optimization of CNN in memristor crossbar.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“However, most of those studies in fact only measured the figures-of-merit of one/few devices and simulated the accuracy of an ANN via software62–67 in such type of studies the connection between the memristors fabricated and the ANN is relatively weak.”
From Hardware implementation of memristor-based artificial neural networks (2024) · cites this paper as [66]
In the citing paper
“some articles focused on the fabrication and device-level characterization of one/few memristors62–70,186,187, also present some of the figures-of-merit generated by a software-based training ANN process (similar to the ones in Fig. 14) in order to claim that their devices exhibit potential for neuromorphic applications.”
From Hardware implementation of memristor-based artificial neural networks (2024) · cites this paper as [66]
Citations
Cites
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 2×
“The rectified linear unit (ReLU) activation function and stochastic gradient descent algorithm are applied [23].”
BackgroundMethodRepresentation Learning: A Review and New Perspectives
Bengio, Y., Courville, A., & Vincent, P. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.50 · Cited by 13,001
Fukushima, K. (1980) · Biological Cybernetics · doi:10.1007/bf00344251 · Cited by 5,759 · cited 1×
“for neuromorphic computing have been demonstrated in recent years, such as multilayer perception (MLP) network [1], convolutional neural network (CNN) [5], spiking neural network (SNN) [6], and long short-term memory (LSTM) network [7].”
BackgroundDeep Convolutional Neural Networks for Image Classification: A Comprehensive Review
Rawat, W., & Wang, Z. (2017) · Neural Computation · doi:10.1162/neco_a_00990 · Cited by 3,503 · cited 1×
“image recognition, classification, and detection tasks [8].”
Training and operation of an integrated neuromorphic network based on metal-oxide memristors
Prezioso, M., Merrikh-Bayat, F., Hoskins, B. D., Adam, G. C., Likharev, K. K., & Strukov, D. B. (2015) · Nature · doi:10.1038/nature14441 · Cited by 3,001
Analogue signal and image processing with large memristor crossbars
Li, C., Hu, M., Li, Y., Jiang, H., Ge, N., Montgomery, E., Zhang, J., Song, W., Dávila, N., Graves, C. E., et al. (2017) · Nature Electronics · doi:10.1038/s41928-017-0002-z · Cited by 1,292 · cited 1×
“2(a)] [27], [28].”
Self-citationBurr, G. W., Shelby, R. M., Sidler, S., di Nolfo, C., Jang, J., Boybat, I., Shenoy, R. S., Narayanan, P., Virwani, K., Giacometti, E. U., et al. (2015) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2015.2439635 · Cited by 936
Face classification using electronic synapses
Yao, P., Wu, H., Gao, B., Eryilmaz, S. B., Huang, X., Zhang, W., Zhang, Q., Deng, N., Shi, L., Wong, H.-S. P., et al. (2017) · Nature Communications · doi:10.1038/ncomms15199 · Cited by 909 · cited 6×
“Here, the conductance range was normalized to [0, 1], therefore, the initial values of W+ and W− are randomly distributed from 0 to 1 (the range of W is [−1, 1]).”
BackgroundInfluentialMemristor‐Based Analog Computation and Neural Network Classification with a Dot Product Engine
Hu, M., Graves, C. E., Li, C., Li, Y., Ge, N., Montgomery, E., Davila, N., Jiang, H., Williams, R. S., Yang, J. J., et al. (2018) · Advanced Materials · doi:10.1002/adma.201705914 · Cited by 751 · cited 1×
“2(a)] [27], [28].”
Self-citationFerroelectric FET analog synapse for acceleration of deep neural network training
Jerry, M., Chen, P.-Y., Zhang, J., Sharma, P., Ni, K., Yu, S., & Datta, S. (2017) · 2017 IEEE International Electron Devices Meeting (IEDM) · doi:10.1109/iedm.2017.8268338 · Cited by 599
Cited by
Hardware implementation of memristor-based artificial neural networks
Aguirre, F., Sebastian, A., Le Gallo, M., Song, W., Wang, T., Yang, J. J., Lu, W., Chang, M.-F., Ielmini, D., Yang, Y., et al. (2024) · Nature Communications · doi:10.1038/s41467-024-45670-9 · Cited by 564
A review of memristor: material and structure design, device performance, applications and prospects
Xiao, Y., Jiang, B., Zhang, Z., Ke, S., Jin, Y., Wen, X., & Ye, C. (2023) · Science and Technology of Advanced Materials · doi:10.1080/14686996.2022.2162323 · Cited by 295
“ing in the switching layer [177], design double layer structure [178,179] have been taken by some academics.”
BackgroundAmirsoleimani, A., Alibart, F., Yon, V., Xu, J., Pazhouhandeh, M. R., Ecoffey, S., Beilliard, Y., Genov, R., & Drouin, D. (2020) · Advanced Intelligent Systems · doi:10.1002/aisy.202000115 · Cited by 216
Multiply accumulate operations in memristor crossbar arrays for analog computing
Chen, J., Li, J., Li, Y., & Miao, X. (2021) · Journal of Semiconductors · doi:10.1088/1674-4926/42/1/013104 · Cited by 109
Self-citationTeixeira, H., Dias, C., Silva, A., & Ventura, J. O. (2024) · ACS Nano · doi:10.1021/acsnano.4c03264 · Cited by 75
Novel Analog Implementation of a Hyperbolic Tangent Neuron in Artificial Neural Networks
Shakiba, F. M., & Zhou, M. (2020) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2020.3034856 · Cited by 79
HfO x /AlO y Superlattice‐Like Memristive Synapse
Wang, C., Mao, G., Huang, M., Huang, E., Zhang, Z., Yuan, J., Cheng, W., Xue, K., Wang, X., & Miao, X. (2022) · Advanced Science · doi:10.1002/advs.202201446 · Cited by 91
Self-citationInterpretable Memristive LSTM Network Design for Probabilistic Residential Load Forecasting
Li, C., Dong, Z. Y., Ding, L., Petersen, H., Qiu, Z., Chen, G., & Prasad, D. K. (2022) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2022.3155443 · Cited by 68
Toward memristive in-memory computing: principles and applications
Bao, H., Zhou, H., Li, J., Pei, H.-Z., Tian, J., Yang, L., Ren, S., Tong, S.-Q., Li, Y., He, Y., et al. (2022) · Frontiers of Optoelectronics · doi:10.1007/s12200-022-00025-4 · Cited by 79
“As for the learning applications, it is well recognized that high precision and stability of the weight update are extremely critical during the network training, demanding large numbers of tunable conductance states, low cycle-to-cycle variation, as well as high symmetricity and linearity in the potentiation and supp…”
BackgroundSelf-citationQin, F. H., Zhang, Y., Song, H. W., & Lee, S. (2023) · Materials Advances · doi:10.1039/d3ma00069a · Cited by 68
2D-Material-Based Volatile and Nonvolatile Memristive Devices for Neuromorphic Computing
Xia, X., Huang, W., Hang, P., Guo, T., Yan, Y., Yang, J., Yang, D., Yu, X., & Li, X. (2023) · ACS Materials Letters · doi:10.1021/acsmaterialslett.2c01026 · Cited by 59
Raikar, A. S., Andrew, J., Dessai, P. P., Prabhu, S. M., Jathar, S., Prabhu, A., Naik, M. B., & Raikar, G. V. S. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10948-3 · Cited by 53
“Grasping these fundamental ethical concepts is imperative before delving into the ethical implications of neuromorphic research (Zendrikov et al. 2023; Chen et al. 2020).”
Zhang, M., Chen, X., Chen, Z., Dan, R., Wei, Y., Rong, H., Wang, Q., Chen, X., Han, A., Wang, Y., et al. (2022) · Applied Surface Science · doi:10.1016/j.apsusc.2022.155956 · Cited by 37
Low Energy Reduction Technique via Memristor for Wireless Body Sensors
Ramesh, K. K. S., Parasuraman, S., Ramesh, G. P., & Rachana, P. G. (2022) · Lecture notes in electrical engineering · doi:10.1007/978-981-19-2281-7_54 · Cited by 30
Recent Progress on Memristive Convolutional Neural Networks for Edge Intelligence
Qin, Y., Bao, H., Wang, F., Chen, J., Li, Y., & Miao, X. (2020) · Advanced Intelligent Systems · doi:10.1002/aisy.202000114 · Cited by 40
Self-citationLee, H., Ryu, D. G., Lee, G., Song, M., Moon, H., Lee, J., Ryu, J., Kang, J., Suh, J. M., Kim, S., et al. (2022) · Advanced Electronic Materials · doi:10.1002/aelm.202200378 · Cited by 30
Kang, Y., Jang, J., Cha, D., & Lee, S. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.3047963 · Cited by 28
Chen, Z., Zhang, J., Wen, S., Li, Y., & Hong, Q. (2021) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2021.3069221 · Cited by 28
“Some studies have quantified several nonideal properties of memristor synaptic devices, such as finite conductance state, writing nonlinearity, and mutation, and comprehensively studied their effects on the performance of CNNs [20].”
BackgroundSelf-citationKim, M., & Kim, S. (2024) · ACS Applied Electronic Materials · doi:10.1021/acsaelm.4c00428 · Cited by 29
Wang, Y., Guo, D., Jiang, J., Wang, H., Shang, Y., Zheng, J., Huang, R., Li, W., & Wang, S. (2024) · ACS Applied Materials & Interfaces · doi:10.1021/acsami.3c18053 · Cited by 24
Dealing with Non-Idealities in Memristor Based Computation-In-Memory Designs
Gebregiorgis, A., Singh, A., Diware, S., Bishnoi, R., & Hamdioui, S. (2022) · IFIP/IEEE International Conference on Very Large Scale Integration · doi:10.1109/vlsi-soc54400.2022.9939618 · Cited by 31
“This work was supported in part by the EU H2020 grant “DAIS” that has received funding from the ECSEL Joint Undertaking (JU) under grant agreement No 101007273 utilize redundancy schemes to tackle the impact of nonidealities [14, 15].”
BackgroundInfluence of Al2O3 layer on InGaZnO memristor crossbar array for neuromorphic applications
Choi, W. S., Jang, J. T., Kim, D., Yang, T. J., Kim, C., Kim, H., & Kim, D. H. (2022) · Chaos Solitons & Fractals · doi:10.1016/j.chaos.2022.111813 · Cited by 26
Naive Bayes classifier based on memristor nonlinear conductance
Li, L., Zhou, Z., Bai, N., Wang, T., Xue, K., Sun, H., He, Q., Cheng, W., & Miao, X. (2022) · Microelectronics Journal · doi:10.1016/j.mejo.2022.105574 · Cited by 28
Self-citationDeep reservoir computing based on self-rectifying memristor synapse for time series prediction
Wang, R., Qi, L., Wang, S., Cao, Y., Ma, X., Wang, H., & Hao, Y. (2023) · Applied Physics Letters · doi:10.1063/5.0158076 · Cited by 22
Zhou, Y., Hu, X., Wang, L., Zhou, G., & Duan, S. (2021) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2021.3115787 · Cited by 23
Threshold learning algorithm for memristive neural network with binary switching behavior
Youn, S., Hwang, Y., Kim, T., Kim, S., Hwang, H., Park, J., & Kim, H. (2024) · Neural Networks · doi:10.1016/j.neunet.2024.106355 · Cited by 19
Brain-like synaptic memristor based on lithium-doped silicate for neuromorphic computing
Ke, S., Jiang, L., Zhao, Y., Xiao, Y., Jiang, B., Cheng, G., Wu, F., Cao, G., Peng, Z., Zhu, M., et al. (2022) · Frontiers of Physics · doi:10.1007/s11467-022-1173-2 · Cited by 18
“The PPF ratio can be defined as follows: PPF ratio = p2 − p1 p1 × 100%, (4) p1 p2in which and are the response current values of the first pulse and the second pulse.”
ResultInfluentialMemristive Fast-Canny Operation for Edge Detection
Tian, J., Zhou, H., Bao, H., Chen, J., Huang, X., Li, J.-C., Yang, L., Li, Y., & Miao, X. (2022) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2022.3204525 · Cited by 17
“Conductance variation of the memristor usually exhibits a Gaussian distribution under the input of consecutive read pulses [30].”
BackgroundSelf-citationMemristor-Based CNNs for Detecting Stress Using Brain Imaging Signals
Bak, S., Park, J., Lee, J., & Jeong, J. (2023) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2023.3297841 · Cited by 16
“It also limits the weight range within ± 1 [47].”
BackgroundResultInfluentialResistive switching memory for high density storage and computing*
Xu, X., Luo, Q., Gong, T., Lv, H., Liu, Q., & Liu, M. (2021) · Chinese Physics B · doi:10.1088/1674-1056/abe0c4 · Cited by 15
Overview of amorphous carbon memristor device, modeling, and applications for neuromorphic computing
Wu, J., Yang, X., Chen, J., Li, S., Zhou, T., Cai, Z., Lian, X., & Wang, L. (2024) · Nanotechnology Reviews · doi:10.1515/ntrev-2023-0181 · Cited by 16
Self-citationExploiting Non-idealities of Resistive Switching Memories for Efficient Machine Learning
Yon, V., Amirsoleimani, A., Alibart, F., Melko, R. G., Drouin, D., & Beilliard, Y. (2022) · Frontiers in Electronics · doi:10.3389/felec.2022.825077 · Cited by 18
“A write-verify scheme (Papandreou et al., 2011; Yi et al., 2011; Alibart et al., 2012; Perez et al., 2017; Pan et al., 2020) typically neglects energy consumption and memory lifetime for the sake of accuracy.”
BackgroundMethodInfluentialJiang, S., Peng, L., Hao, Z., Du, X., Gu, J., Su, J., Guo, H., Gu, D., Zhang, H., Wang, Q., et al. (2023) · ACS Applied Electronic Materials · doi:10.1021/acsaelm.3c00698 · Cited by 13
Self-citationConfigurable Kernel Map Implementation in Memristor Crossbar for Convolution Neural Network
Kim, G., Youn, S., Park, J., & Kim, H. (2025) · Advanced Intelligent Systems · doi:10.1002/aisy.202500032 · Cited by 14
Koh, E. K., Dananjaya, P. A., Liu, L., Lee, C. X. X., Lim, G. J., You, Y. S., & Lew, W. S. (2024) · ACS Nano · doi:10.1021/acsnano.4c07454 · Cited by 12
RRAM Crossbar-Based Fault-Tolerant Binary Neural Networks (BNNs)
Gebregiorgis, A., Zografou, A., & Hamdioui, S. (2022) · IEEE European Test Symposium (ETS) · doi:10.1109/ets54262.2022.9810414 · Cited by 9
“Some of them focus on optimal mapping [7], [4], while others rely on retraining [5], [6].”
BackgroundMethodZhang, Y., Ma, Z., Chen, Z., Poddar, S., Zhu, Y., Han, B., Chan, C. L. J., Ding, Y., Kong, X., & Fan, Z. (2024) · Advanced Intelligent Systems · doi:10.1002/aisy.202300586 · Cited by 9
Simulation of In-Situ Training in Spike Neural Network Based on Non-Ideal Memristors
Ma, L., Wang, G., Wang, S., & Chen, D. (2023) · IEEE Journal of the Electron Devices Society · doi:10.1109/jeds.2023.3311763 · Cited by 9
“Furthermore, some works have made great progress on the asymmetric nonlinear update weights of RRAM in ANNs, [12], [13], [14], [15], [16], [17] but for the SNNs, there is only few research on no-ideal devices.”
BackgroundArtificial synaptic properties of zirconium-doped barium titanate film for neuromorphic computing
Zheng, Y.-F., Tang, X., Fang, J., Lai, X.-C., Jiang, Y., & Liu, Q. (2024) · Materials Today Communications · doi:10.1016/j.mtcomm.2024.108893 · Cited by 8
Wu, Z., Li, Z., Lin, X., Shan, X., Chen, G., Yang, C., Zhao, X., Sun, Z., Hu, K., Wang, F., et al. (2023) · Nanotechnology · doi:10.1088/1361-6528/acf0c8 · Cited by 8
“Furthermore, LiSiO x -based memristors have also been developed for neuromorphic computing applications due to their superior performance, such as multilayer perceptron [ 24, 25 ] , convolutional neural networks [ 26 ] , spiking neural networks [ 27, 28 ] .”
On the Reliability of RRAM-Based Neural Networks
Aziza, H., Zambelli, C., Hamdioui, S., Diware, S., Bishnoi, R., & Gebregiorgis, A. (2023) · IFIP/IEEE International Conference on Very Large Scale Integration · doi:10.1109/vlsi-soc57769.2023.10321859 · Cited by 11
“Similarly, the hardware-based solutions utilize redundancy schemes to tackle the impact of non-idealities [19], [20].”
BackgroundJiang, L., Jin, Y., Zhao, Y., Meng, J., Zhang, J., Chen, X., Wu, X., Xiao, Y., Tao, Z., Jiang, B., et al. (2023) · Advanced Physics Research · doi:10.1002/apxr.202200086 · Cited by 10
Yin, S.-F., Sun, Q., Liu, L.-F., Liu, S.-Z., Jiang, Y., & Tang, X. (2025) · Applied Surface Science · doi:10.1016/j.apsusc.2025.164049 · Cited by 8
LATIM: Loading-Aware Offline Training Method for Inverter-Based Memristive Neural Networks
Vahdat, S., Kamal, M., Afzali‐Kusha, A., & Pedram, M. (2021) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2021.3072289 · Cited by 7
Xiao, Y., Ke, S., Jin, Y., Zhang, T., Jiang, B., Meng, J., Chen, S., Li, R., Zhang, Z., & Ye, C. (2023) · Ceramics International · doi:10.1016/j.ceramint.2023.12.211 · Cited by 10
High-Linearity Ta2O5 Memristor and Its Application in Gaussian Convolution Image Denoising
Wang, Y., Wang, H., Guo, D., An, Z., Zheng, J., Huang, R., Bi, A., Jiang, J., & Wang, S. (2024) · ACS Applied Materials & Interfaces · doi:10.1021/acsami.4c09056 · Cited by 6
Study on Memristor-Based 3-D Convolutional Neural Network Adapting to Non-Ideal Effects
Dong, B., Li, P., Fu, Y., Chang, T., Li, Y., Miao, X., & He, Y. (2023) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2022.3233552 · Cited by 6
“Especially, previous work mainly focused on the evaluation of non-ideal effects, while did not make further analysis and optimization based on the evaluation results [11], [12].”
MethodSelf-citationHan, J., Park, J., Rehman, S., Khan, M. F., Kim, M., & Kim, S. (2025) · InfoMat · doi:10.1002/inf2.70018 · Cited by 6
Tong, Z.-L., Jiang, Y. L., Min-Huang, Tang, X., Tang, Z., Guo, X., Li, W., & Zhou, Y. (2025) · Physica B Condensed Matter · doi:10.1016/j.physb.2025.417412 · Cited by 5
Prabhu, N. L., & Raghavan, N. (2022) · IEEE Access · doi:10.1109/access.2022.3219066 · Cited by 5
“The studies on the re-programming and verification using various programming architectures involving RRAM device model data [16], [17], [18], [19], [20] or actual fabricated device array data or intelligent workload mapping devices array data [21], [22], [23], [24], [25] are discussed in Table 2.”
MethodZhang, Y., Sun, H., Xie, M., Feng, Z., & Wu, Z. (2023) · Advanced Intelligent Systems · doi:10.1002/aisy.202300411 · Cited by 5
Nb 2 O 5 memristive neurons‐based unsupervised learning network
Lu, Z., Zhu, Q., Shi, S., Wang, K., & Liang, Y. (2024) · International Journal of Circuit Theory and Applications · doi:10.1002/cta.4044 · Cited by 5
“11,12 13 – 15 Recently, researchers have used memristors as synapses to realize Hopfield network, 16,17 CNN network, 18,19 spike neural network, 5,20,21 and so on.”
Wu, Q., Xie, C., Li, X., Chen, H., Xie, L., Wang, Q., Sui, Z., Song, S., & Song, Z. (2023) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2023.3288514 · Cited by 5
“A weight pruning method, proposed in [21] and [22] to limit the weight range, was applied to prune the weights within a certain range.”
MethodAsanuma, S., Sumita, K., Miyaguchi, Y., Horita, K., Masuda, T., Jimbo, T., & Miyata, N. (2023) · Applied Physics Express · doi:10.35848/1882-0786/acdd4e · Cited by 4
Lee, J., Hwang, J., Cho, Y., Park, M., Choi, W. Y., Kim, S., & Lee, J. (2022) · IEEE Journal on Exploratory Solid-State Computational Devices and Circuits · doi:10.1109/jxcdc.2022.3220032 · Cited by 8
“They also necessitate a specific pulse generator design or precise conductance sensing to guarantee an acceptable level of accuracy [20], [22].”
BackgroundMiyata, N. (2024) · Electronics · doi:10.3390/electronics13040726 · Cited by 3
Loading-Aware Reliability Improvement of Ultra-Low Power Memristive Neural Networks
Vahdat, S., Kamal, M., Afzali‐Kusha, A., & Pedram, M. (2021) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2021.3084867 · Cited by 4
NMBNN: Noise-Adaptive Memristive Bayesian Neural Network for Energy-Efficient Edge Health Care
Li, H., Tian, F., Yang, J., Sawan, M., & El‐Atab, N. (2023) · IEEE Biomedical Circuits and Systems Conference (BioCAS) · doi:10.1109/biocas58349.2023.10388528 · Cited by 4
“Those variations generate notable deviations from the expected weight value and result in measurement errors, which degrade the performance in conventional neural networks [12].”
BackgroundBao, H., Qin, Y., Chen, J., Yang, L., Li, J., Zhou, H., Li, Y., & Miao, X. (2022) · Frontiers in Electronics · doi:10.3389/felec.2022.954661 · Cited by 3
“According to the previous research (Pan et al., 2020), the weight parameters of the networks are centralized around zeros and can be clipped within [ − 0.25, 0.25] during quantization to obtain better network accuracies.”
BackgroundMethodInfluentialSelf-citationGuo, Z., Zhang, Y., Xu, S., Wu, Z., & Zhao, W. (2021) · IEEE International Conference on ASIC (ASICON) · doi:10.1109/asicon52560.2021.9620424 · Cited by 2
Genetic Algorithm-Based Training Method for Memristor Neural Networks
Zhang, W., Zhang, Q., & Wu, H. (2024) · International Journal of Computer Applications · doi:10.5120/ijca2024924139 · Cited by 2
“[9 ] employed committee machines to improve inference accuracy through ensemble averaging, and [17] introduced methods to alleviate the impacts of conductance state asymmetry and cycle-to-cycle variations.”
Han, C., Zhang, M., Liu, Y., Wang, X., Zhang, H., & Tong, Y. (2025) · IEEE Electron Devices Technology & Manufacturing Conference (EDTM) · doi:10.1109/edtm61175.2025.11041024 · Cited by 1
High-accuracy inference using HfOxSy/HfS2 memristors
Xhameni, A., & Lombardo, A. (2025) · APL Electronic Devices · doi:10.1063/5.0286727 · Cited by 1
“Pan et al., 21 measure devices with a similar structure to ours (TiN/HfO 2 /Ti), and Yao et al., also use a hafnia-based device stack (TiN/TaO x /HfO x /TiN).”
Evolution of Neural Network Models and Computing-in-Memory Architectures
Kumar, A., Devi, M., Singh, S. K., Singh, K., & Mishra, D. (2025) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-025-10481-8 · Cited by 3
“Wen-Qian Pan et al. [ 47 proposed three strategies to mitigate these issues: (1) limiting the weight range to enhance N-state utilization; (2) adopting a “with-read” update scheme to address asymmetric write nonlinearities; and (3) using multiple memristors per kernel element to reduce the impact…”
Variation-Tolerant Circuit Design and Online Learning Framework for Memristor-Based Trace-STDP SNN
Zheng, Y., He, Y., Shen, R., Li, J., Wang, R., Wang, D., Zheng, L., Zou, Z., & Xu, J. (2026) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2026.3663895 · Cited by 1
Self-citationAnalysis of Memristive Quantized Convolutional Neural Network Accelerator with Device Nonideality
Bao, H., Huang, X., Chen, J., Yi, L., & Miao, X. (2020) · IEEE International Conference on Integrated Circuits, Technologies and Applications (ICTA) · doi:10.1109/icta50426.2020.9332143
“Truncated range of (-0.25, 0.25) is used to compensate the effect of the limited precision for best performance, as previously discussed in [8], In the forward propagation, the full-precision weights are quantized before used in the network to generate output and loss.”
MethodInfluentialSelf-citationDesign for Enhancing the Performance of Memristive Convolution Neural Network
Chen, J., Pan, W.-O., Li, Y., Chang, T., & Miao, X. (2020) · IEEE International Conference on Integrated Circuits, Technologies and Applications (ICTA) · doi:10.1109/icta50426.2020.9332060
Self-citationLee, J., Hwang, J., Cho, Y., Kim, S., & Lee, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.07113 · Cited by 1
“They also sacrifice sensing precision to achieve ~90 % accuracy on the MNIST dataset under high NL settings [21,23].”
BackgroundMethodSasikumar, R., & Padmanabhan, R. (2023) · Conference on Ph.D. Research in Microelectronics and Electronics · doi:10.1109/prime58259.2023.10161791
“A combined software-hardware approach has been proposed to analyze the system performance of memristor-based neural networks [4], [5], [7], [8], [10], [11].”
BackgroundMethodBITLITE: Light Bit-wise Operative Vector Matrix Multiplication for Low-Resolution Platforms
Tran, V., Chen, D., Genov, R., Azghadi, M. R., & Amirsoleimani, A. (2024) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas58744.2024.10558615
“However, memristors have finite intermediate conductance stages and are susceptible to drift, read noise, and asymmetry due to inherent device imperfections [14]–[16], which in turn, constrain the performance of ML algorithms on crossbar circuits.”
BackgroundX-IMM: Mixed-Signal Iterative Montgomery Modular Multiplication
Kamal, M., & Pedram, M. (2024) · International Symposium on Low Power Electronics and Design (ISLPED) · doi:10.1145/3665314.3670849
Heo, D., Bang, H., & Ko, J. H. (2024) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1145/3649329.3656535
“Also, C2C variations introduce immediate conductance distortions on application of voltage pulses [16], while D2D variations influence nonlinearity of NDs.”
Yin, S., Song, H., Ma, C., Yan, J., Lin, C., Wang, J., & Zhong, X. (2026) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2026.3663336
“The memristor’s conductance update characteristics significantly impact the computational performance of memristive neural network hardware [16], [17].”
RRAM co-integration with OTS-based back-end selector : from devices to applications
Lopez, J. M. (2022) · theses.fr (ABES) · doi:10.70675/197a0071z4b82z48fdzad97zce487a71c1af
Sun, K., Liu, H., Yang, Y., & Zuo, Z. (2026) · Nanotechnology · doi:10.1088/1361-6528/ae6f23
“Current research focuses on multifunctional integration and energy efficiency, requiring synergistic development of machine learning algorithms (e.g. convolutional neural networks (CNNs) [9, 10], reservoir computing (RC) [11]) and heterogeneous [12] hardware systems.”
Metadata
- Authors
- Pan, W.-Q., Chen, J., Kuang, R., Li, Y., He, Y.-H., Feng, G.-R., Duan, N., Chang, T.-C., & Miao, X.-S.
- Affiliations
- Wuhan National Laboratory for Optoelectronics (CN)
- National Sun Yat-sen University (TW)
- Venue
- IEEE Transactions on Electron Devices
- Published
- 2020-03
- Volume
- 67
- Issue
- 3
- Pages
- 895-901
- ISSN
- 0018-9383 (ISSN-L)
- 1557-9646
- OpenAlex
- W3001813906
- Cited by
- 76
- References
- 41
- Citations by year
- 2026: 4 · 2025: 7 · 2024: 17 · 2023: 17 · 2022: 17 · 2021: 9 · 2020: 5
- FWCI
- 4.11
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- Locations
- IEEE Transactions on Electron Devices (publishedVersion)
- Funders
- National Key Research and Development Plan of Ministry of Science and Technology (MOST) of China (2016YFA0203800)
- National Natural Science Foundation of China (61841404, 51732003, 61674061, 61774067)
- National Defense Advanced Research Program (31513040304)
- Major Special Projects for Technological Innovation of Hubei Province (2018AAA038)
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Ferroelectric and Negative Capacitance Devices
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- memristors, convolutional neural networks, memristor crossbar, artificial synapses, conductance states, cycle-to-cycle variation, device-to-device variation, learning accuracy
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.