Inference of Long-Short Term Memory networks at software-equivalent accuracy using 2.5M analog Phase Change Memory devices
The DOI matches its Crossref record; 4 of the 4 references it cites were found in a registry or an index.
Abstract
We report accuracy for forward inference of long-short-term-memory (LSTM) networks using weights programmed into the conductances of $> 2.5\text{M}$ phase-change memory (PCM) devices. We demonstrate strategies for software weight-mapping and programming of hardware analog conductances that provide accurate weight programming despite significant device variability. Inference accuracy very close to software-model baselines is achieved on several language modeling tasks.
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
“For all levels, we achieve a standard deviation less than 1.2 μS, which is more than two times lower than that reported in previous works on nanoscale PCM arrays for a similar conductance range40,41.”
From Accurate deep neural network inference using computational phase-change memory (2020) · cites this paper as [41]
In the citing paper
“These issues cause both MVM read-out31 and the writing or programming of the conductances32–34 to be erroneous and non-deterministic.”
From Hardware-aware training for large-scale and diverse deep learning inference workloads using in-memory computing-based accelerators (2023) · cites this paper as [33]
In the citing paper
“A number of previous studies have shown that noise-aware training —simple injection of noise onto weights or activations during DNN training—can make DNNs more robust for AIMC deployment33,38–42.”
From Hardware-aware training for large-scale and diverse deep learning inference workloads using in-memory computing-based accelerators (2023) · cites this paper as [33]
Citations
Cites
Equivalent-accuracy accelerated neural-network training using analogue memory
Ambrogio, S., Narayanan, P., Tsai, H., Shelby, R. M., Boybat, I., di Nolfo, C., Sidler, S., Giordano, M., Bodini, M., Farinha, N. C. P., et al. (2018) · Nature · doi:10.1038/s41586-018-0180-5 · Cited by 1,151
Self-citationGokmen, T., & Vlasov, Y. (2016) · Frontiers in Neuroscience · doi:10.3389/fnins.2016.00333 · Cited by 491
Long short-term memory networks in memristor crossbar arrays
Li, C., Wang, Z., Rao, M., Belkin, D., Song, W., Jiang, H., Yan, P., Li, Y., Lin, P., Hu, M., et al. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-018-0001-4 · Cited by 429
Training LSTM Networks With Resistive Cross-Point Devices
Gokmen, T., Rasch, M. J., & Haensch, W. (2018) · Frontiers in Neuroscience · doi:10.3389/fnins.2018.00745 · Cited by 53
Research works printed but not linked: 4
- giannopoulos, 2018, IEDM, 27
- mackin, 2019, Adv Electron Mater
- kim, 2013, IEDM, 30
- burr, 2014, IEDM, t29 5
Cited by
Resistive switching materials for information processing
Wang, Z., Wu, H., Burr, G. W., Hwang, C. S., Wang, K. L., Xia, Q., & Yang, J. J. (2020) · Nature Reviews Materials · doi:10.1038/s41578-019-0159-3 · Cited by 1,214
Self-citationAccurate deep neural network inference using computational phase-change memory
Joshi, V., Le Gallo, M., Haefeli, S., Boybat, I., Nandakumar, S. R., Piveteau, C., Dazzi, M., Rajendran, B., Sebastian, A., & Eleftheriou, E. (2020) · Nature Communications · doi:10.1038/s41467-020-16108-9 · Cited by 490
Accurate deep neural network inference using computational phase-change memory
Joshi, V., Gallo, M. L., Haefeli, S., Boybat, I., Nandakumar, S. R., Piveteau, C., Dazzi, M., Rajendran, B., Sebastian, A., & Eleftheriou, E. S. (2020) · RePEc: Research Papers in Economics · doi:10.3929/ethz-b-000416797 · Cited by 379
Le Gallo, M., Khaddam-Aljameh, R., Stanisavljevic, M., Vasilopoulos, A., Kersting, B., Dazzi, M., Karunaratne, G., Brändli, M., Singh, A., Müller, S. M., et al. (2023) · Nature Electronics · doi:10.1038/s41928-023-01010-1 · Cited by 310
Analog architectures for neural network acceleration based on non-volatile memory
Xiao, T. P., Bennett, C. H., Feinberg, B., Agarwal, S., & Marinella, M. J. (2020) · Applied Physics Reviews · doi:10.1063/1.5143815 · Cited by 193
Rasch, M. J., Mackin, C., Le Gallo, M., Chen, A., Fasoli, A., Odermatt, F., Li, N., Nandakumar, S. R., Narayanan, P., Tsai, H., et al. (2023) · Nature Communications · doi:10.1038/s41467-023-40770-4 · Cited by 213
“These issues cause both MVM read-out [56] and the writing or programming of the conductances [60, 79, 51] to be erroneous and non-deterministic.”
BackgroundSelf-citationA Flexible and Fast PyTorch Toolkit for Simulating Training and Inference on Analog Crossbar Arrays
Rasch, M. J., Moreda, D., Gokmen, T., Le Gallo, M., Carta, F., Goldberg, C., El Maghraoui, K., Sebastian, A., & Narayanan, V. (2021) · 2021 IEEE 3rd International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas51828.2021.9458494 · Cited by 160
“It has become increasingly apparent that accuracy on chips that are designed for inference benefit greatly from “hardware-aware” training, where the DNN workload is trained (or fine-tuned) in software with some analog noise and nonidealities in the forward pass, but with otherwise perfect backward and update pass [4,8…”
BackgroundBonnet, D., Hirtzlin, T., Majumdar, A., Dalgaty, T., Esmanhotto, E., Meli, V., Castellani, N., Martin, S., Nodin, J.-F., Bourgeois, G., et al. (2023) · Nature Communications · doi:10.1038/s41467-023-43317-9 · Cited by 108
“In phase-change memories, the trend is inverted: device variability increases with the conductance values, moving from a full to a partial amorphous material 24,25 .”
Using the IBM analog in-memory hardware acceleration kit for neural network training and inference
Gallo, M. L., Lammie, C., Büchel, J., Carta, F., Fagbohungbe, O. I., Mackin, C., Tsai, H.-Y. S., Narayanan, V., Sebastian, A., Maghraoui, K. E., et al. (2023) · APL Machine Learning · doi:10.1063/5.0168089 · Cited by 100
Self-citationChallenges and Trends of Nonvolatile In-Memory-Computation Circuits for AI Edge Devices
Hung, J.-M., Jhang, C.-J., Wu, P.-C., Chiu, Y.-C., & Chang, M. (2021) · IEEE Open Journal of the Solid-State Circuits Society · doi:10.1109/ojsscs.2021.3123287 · Cited by 65
Hung, J.-M., Li, X., Wu, J., & Chang, M. (2020) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2020.2976115 · Cited by 56
“Tsai et al. [38] demonstrated a long-short-term-memory (LSTM) neural network using up to 4-”
MethodNoise-Resilient DNN: Tolerating Noise in PCM-Based AI Accelerators via Noise-Aware Training
Kariyappa, S., Tsai, H., Spoon, K., Ambrogio, S., Narayanan, P., Mackin, C., Chen, A., Qureshi, M., & Burr, G. W. (2021) · IEEE Transactions on Electron Devices · doi:10.1109/ted.2021.3089987 · Cited by 45
Self-citationRoadmap on low-power electronics
Ramesh, R., Salahuddin, S., Datta, S., Diaz, C. H., Nikonov, D. E., Young, I. A., Ham, D., Chang, M., Khwa, W.-S., Lele, A. S., et al. (2024) · APL Materials · doi:10.1063/5.0184774 · Cited by 47
Federated Learning for Supply Chain Demand Forecasting
Wang, H., Xie, F., Duan, Q., & Li, J. (2022) · Mathematical Problems in Engineering · doi:10.1155/2022/4109070 · Cited by 42
A near-threshold memristive computing-in-memory engine for edge intelligence
Wang, L., Li, W., Zhou, Z., An, J., Ye, W., Li, Z., Gao, H., Hu, H., Liu, J., Chen, X., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-61025-4 · Cited by 38
Spoon, K., Tsai, H., Chen, A., Rasch, M. J., Ambrogio, S., Mackin, C., Fasoli, A., Friz, A. M., Narayanan, P., Stanisavljevic, M., et al. (2021) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2021.675741 · Cited by 26
“However, NVM devices exhibit many conductance instabilities [conductance drift (Ambrogio et al., 2019), programming and read noise (Tsai et al., 2019), etc.], which can degrade accuracy, particularly as the time between programming and inference increases.”
BackgroundSelf-citationYang, J., Mao, R., Jiang, M., Cheng, Y., Sun, P.-S. V., Dong, S., Pedretti, G., Sheng, X., Ignowski, J., Li, H., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-56254-6 · Cited by 38
Memristive System Based Image Processing Technology: A Review and Perspective
Ji, X., Dong, Z., Zhou, G., Lai, C. S., Yan, Y., & Qi, D. (2021) · Electronics · doi:10.3390/electronics10243176 · Cited by 29
“[73] reported a long short-term memory network, which mapped and programmed the network weights into the phase-change memory devices, as demonstrated in Figure 19.”
BackgroundSelf-Aware Neural Network Systems: A Survey and New Perspective
Du, Z., Guo, Q., Zhao, Y., Zhi, T., Chen, Y., & Xu, Z. (2020) · Proceedings of the IEEE · doi:10.1109/jproc.2020.2977722 · Cited by 21
Evolving conductive polymer neural networks on wetware
Akai‐Kasaya, M., Hagiwara, N., Hikita, W., Okada, M., Sugito, Y., Kuwahara, Y., & Asai, T. (2020) · Japanese Journal of Applied Physics · doi:10.35848/1347-4065/ab8e06 · Cited by 23
Das, S., Chen, A., & Marinella, M. J. (2021) · doi:10.1109/irds54852.2021.00011 · Cited by 30
Self-citationALPINE: Analog In-Memory Acceleration with Tight Processor Integration for Deep Learning
Klein, J. A. H., Boybat, I., Qureshi, Y. M., Dazzi, M., Levisse, A., Ansaloni, G., Zapater, M., Sebastian, A., & Atienza, D. (2022) · IEEE Transactions on Computers · doi:10.1109/tc.2022.3230285 · Cited by 27
“In our second exploration, we look at recurrent neural networks (RNNs) in the form of a Long-short term memory LSTM targeting character recognition using the Penn Treebank (PTB) data set [41].”
InfluentialBeyond-CMOS roadmap—from Boolean logic to neuro-inspired computing
Chen, A. (2022) · Japanese Journal of Applied Physics · doi:10.35848/1347-4065/ac5d86 · Cited by 15
Self-citationAccelerating Deep Neural Networks with Analog Memory Devices
Ambrogio, S., Narayanan, P., Tsai, H.-Y. S., Mackin, C., Spoon, K., Chen, A., Fasoli, A., Friz, A. M., & Burr, G. W. (2020) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas48895.2020.9073978 · Cited by 13
Self-citationGeneral Methodology for the Design of Bell-Shaped Analog-Hardware Classifiers
Alimisis, V., Eleftheriou, N. P., Kamperi, A., Gennis, G., Dimas, C., & Sotiriadis, P. P. (2023) · Electronics · doi:10.3390/electronics12204211 · Cited by 11
Accelerating Deep Neural Networks with Analog Memory Devices
Spoon, K., Ambrogio, S., Narayanan, P., Tsai, H.-Y. S., Mackin, C., Chen, A., Fasoli, A., Friz, A. M., & Burr, G. W. (2020) · IEEE International Memory Workshop (IMW) · doi:10.1109/imw48823.2020.9108149 · Cited by 12
Self-citationSystem Design for Computation-in-Memory: From Primitive to Complex Functions
Zahedi, M., Shahroodi, T., Custers, G., Singh, A., Wong, S., & Hamdioui, S. (2022) · IFIP/IEEE International Conference on Very Large Scale Integration · doi:10.1109/vlsi-soc54400.2022.9939571 · Cited by 11
“operation VMM VMM VMM VMM Signal and image processing Compressed sensing [28] Discrete Fourier Transform [29] Convolutional Image filtering Huffman encoding GLCM feature extraction operation VMM VMM VMM Matching Addition Classification and Prediction Recurrent neural network [30] Convolutional neural network [12] Hype…”
MethodHan, J., Rasch, M. J., Liu, Z., Solomon, P. M., Brew, K. W., Cheng, K., Ok, I., Chan, V., Longstreet, M., Kim, W., et al. (2022) · Advanced Intelligent Systems · doi:10.1002/aisy.202100179 · Cited by 9
Enabling High-Performance DNN Inference Accelerators Using Non-Volatile Analog Memory (Invited)
Chen, A., Ambrogio, S., Narayanan, P., Tsai, H.-Y. S., Mackin, C., Spoon, K., Friz, A. M., Fasoli, A., & Burr, G. W. (2020) · IEEE Electron Devices Technology & Manufacturing Conference (EDTM) · doi:10.1109/edtm47692.2020.9117896 · Cited by 7
Self-citationA Low-Power Analog Integrated Euclidean Distance Radial Basis Function Classifier
Alimisis, V., Dimas, C., & Sotiriadis, P. P. (2024) · Electronics · doi:10.3390/electronics13050921 · Cited by 9
Hu, Q., Dong, B., Wang, L., Huang, E., Tong, H., He, Y., Xu, M., & Miao, X. (2020) · Chinese Physics B · doi:10.1088/1674-1056/ab892a · Cited by 6
Energy-Efficient Recurrent Neural Network With MRAM-Based Probabilistic Activation Functions
Sheikhfaal, S., Angizi, S., & DeMara, R. F. (2022) · IEEE Transactions on Emerging Topics in Computing · doi:10.1109/tetc.2022.3202112 · Cited by 5
“Generally, ANNs based on von-Neumann architecture face the memory-wall challenge, resulting from the limited memory bandwidth, high energy consumption for data movement between memory and processing units, and long memory access latency [2], [3].”
MethodNeuromorphic Computing with Phase Change, Device Reliability, and Variability Challenges
Mackin, C., Narayanan, P., Ambrogio, S., Tsai, H.-Y. S., Spoon, K., Fasoli, A., Chen, A., Friz, A. M., Shelby, R. M., & Burr, G. W. (2020) · IEEE International Reliability Physics Symposium proceedings · doi:10.1109/irps45951.2020.9128315 · Cited by 6
“Such analog states can be programmed with a row-by-row parallel Closed-Loop Tuning (CLT) procedure (Figure 2a) [24], [34] where after every programming step, the current weight vector is compared with the target vector.”
BackgroundMethodInfluentialSelf-citationInference of Deep Neural Networks with Analog Memory Devices
Ambrogio, S., Narayanan, P., Tsai, H.-Y. S., Mackin, C., Spoon, K., Chen, A., Fasoli, A., Friz, A. M., & Burr, G. W. (2020) · International Symposium on VLSI Technology, Systems and Applications (VLSI-TSA) · doi:10.1109/vlsi-tsa48913.2020.9203663 · Cited by 3
“This increased programming accuracy comes with the penalty of doubling the number of PCM devices [8].”
BackgroundMethodInfluentialSelf-citationNeuromorphic Devices Based on Chalcogenide Materials
Chen, J., Lu, Y., Yang, Z., Li, Y., & Miao, X. (2021) · Wiley eBooks · doi:10.1002/9783527835317.ch4 · Cited by 4
Phase Change Memory-based Hardware Accelerators for Deep Neural Networks (invited)
Burr, G. W., Narayanan, P., Ambrogio, S., Okazaki, A., Tsai, H.-Y. S., Hosokawa, K., Mackin, C., Nomura, A., Yasuda, T., Demarest, J. J., et al. (2023) · IEEE Symposium on VLSI Technology and Circuits (VLSI Technology and Circuits) · doi:10.23919/vlsitechnologyandcir57934.2023.10185411 · Cited by 3
Self-citationFast-Converging Hybrid Pulse for High Precision Multilevel Phase Change Memory
Hong, T., Chen, X., Song, S., Jiao, Z., Li, X., Li, S., Dai, W., & Song, Z. (2025) · IEEE Electron Device Letters · doi:10.1109/led.2025.3633772 · Cited by 1
“Parallel use of multiple PCM devices can improve weight resolution [12], [13], [14]. but at the expense of significant area and power overhead.”
Pruning random resistive memory for optimizing analog AI
Li, Y., Wang, S., Zhao, Y., Wang, S., Wang, B., Zhang, W., He, Y., Lin, N., Cui, B., Chen, X., et al. (2026) · Nature Communications · doi:10.1038/s41467-025-67960-6 · Cited by 6
Methods for Setting Device Specifications for Analog In‐Memory Computing Inference
Wu, Z., Su, X., Rasch, M. J., & Li, N. (2026) · Advanced Intelligent Systems · doi:10.1002/aisy.202501355 · Cited by 1
Kozu, K., & Namba, K. (2021) · IEEE International Conference on Consumer Electronics - Taiwan (ICCE-Taiwan) · doi:10.1109/icce-tw52618.2021.9602979 · Cited by 1
Optimization of Analog Accelerators for Deep Neural Networks Inference
Fasoli, A., Ambrogio, S., Narayanan, P., Tsai, H.-Y. S., Mackin, C., Spoon, K., Friz, A. M., Chen, A., & Burr, G. W. (2020) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas45731.2020.9180799 · Cited by 2
Self-citationHosokawa, K., Narayanan, P., Ambrogio, S., Tsai, H.-Y. S., Mackin, C., Fasoli, A., Friz, A. M., Chen, A., Luquin, J., Spoon, K., et al. (2021) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas51556.2021.9401128 · Cited by 1
Self-citationEvaluating complexity and resilience trade-offs in emerging memory inference machines.
Bennett, C. H., Dellana, R., Xiao, T. P., Feinberg, B., Agarwal, S., Cardwell, S., Marinella, M. J., Severa, W., & Aimone, J. B. (2021) · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · doi:10.2172/1859699
Accelerating Deep Neural Networks with Phase-Change Memory Devices
Spoon, K., Ambrogio, S., Narayanan, P., Tsai, H.-Y. S., Mackin, C., Chen, A., Fasoli, A., Friz, A. M., & Burr, G. W. (2022) · Springer eBooks · doi:10.1007/978-3-031-03841-9_3
Self-citationLi, Y., Liu, Y., Zhou, X., Yang, J., Li, Z., Mei, Y., Yu, W., Zhu, B. H., Wu, X., Ding, S., et al. (2023) · Electronics · doi:10.3390/electronics12163466
Henwood, S., Mordido, G., Savaria, Y., Chandar, S., & Leduc-Primeau, F. (2024) · Asilomar Conference on Signals, Systems and Computers (ACSSC) · doi:10.1109/ieeeconf60004.2024.10943038
“The conductance value that will be seen by the computations will differ from the desired value as a result of programming and read noise [7] as well as conductance drift over time [11].”
Synaptic uncertainty for in-memory computing
Bonnet, D. (2024) · theses.fr (ABES) · doi:10.70675/5d34cb96z0f54z44b0zb3aez48f1fbf1e560
Metadata
- Authors
- Tsai, H., Ambrogio, S., Mackin, C., Narayanan, P., Shelby, R. M., Rocki, K., Chen, A., & Burr, G. W.
- Venue
- 2019 Symposium on VLSI Technology
- Publisher
- IEEE
- Published
- 2019-06
- Pages
- T82-T83
- OpenAlex
- W2964447465
- Cited by
- 47
- References
- 4
- Citations by year
- 2026: 2 · 2025: 3 · 2024: 4 · 2023: 7 · 2022: 7 · 2021: 10 · 2020: 14
- FWCI
- 12.33
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Symposium on VLSI Technology (publishedVersion)
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
- Physical Sciences › Materials Science › Materials Chemistry › Phase-change materials and chalcogenides
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Reservoir Computing
- Keywords
- long short-term memory networks, phase-change memory, analog hardware, inference accuracy, weight mapping, device variability, language modeling, neuromorphic hardware, analog neural networks
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.