The DOI matches its Crossref record; 144 of the 157 references it cites were found in a registry or an index.
Efficient Processing of Deep Neural Networks: A Tutorial and Survey
Abstract
Deep neural networks (DNNs) are currently widely used for many artificial intelligence (AI) applications including computer vision, speech recognition, and robotics. While DNNs deliver state-of-the-art accuracy on many AI tasks, it comes at the cost of high computational complexity. Accordingly, techniques that enable efficient processing of DNNs to improve energy efficiency and throughput without sacrificing application accuracy or increasing hardware cost are critical to the wide deployment of DNNs in AI systems. This article aims to provide a comprehensive tutorial and survey about the recent advances toward the goal of enabling efficient processing of DNNs. Specifically, it will provide an overview of DNNs, discuss various hardware platforms and architectures that support DNNs, and highlight key trends in reducing the computation cost of DNNs either solely via hardware design changes or via joint hardware design and DNN algorithm changes. It will also summarize various development resources that enable researchers and practitioners to quickly get started in this field, and highlight important benchmarking metrics and design considerations that should be used for evaluating the rapidly growing number of DNN hardware designs, optionally including algorithmic codesigns, being proposed in academia and industry. The reader will take away the following concepts from this article: understand the key design considerations for DNNs; be able to evaluate different DNN hardware implementations with benchmarks and comparison metrics; understand the tradeoffs between various hardware architectures and platforms; be able to evaluate the utility of various DNN design techniques for efficient processing; and understand recent implementation trends and opportunities.
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
“Classical arithmetic operations, such as MAC, constitute the bulk of the computational requirements for models used in artificial intelligence [13].”
From Brain-inspired computing systems: a systematic literature review (2024) · cites this paper as [13]
In the citing paper
“To clarify, consider the multiply-and-accumulate (MAC) operation in digital processors, where the operation energy consumption is significantly dependent on the bit width of the input operands [13, 235] or equivalently to the signal-to-noise ratio (SNR) in analog computing [236].”
From Brain-inspired computing systems: a systematic literature review (2024) · cites this paper as [13]
In the citing paper
“Altogether, processing one 224 × 224 input image needs a VGG-16 model of 15.5 G multiply-and-accumulates and 138 M weights [40].”
From Face Recognition Based on Deep Learning and FPGA for Ethnicity Identification (2022) · cites this paper as [40]
In the citing paper
“The energy consumption of executing a DNN model is highly dictated by the computation architecture and methods (e.g., ultra-low power compute-in-memory AI accelerator) at the edge computation nodes [249], the architecture of DNN models [250], and the wireless transmission for data exchange during the model inference procedure.”
From Edge artificial intelligence for 6g: vision, enabling technologies, and applications (2022) · cites this paper as [250]
In the citing paper
“but there is no review discussing all three categories which in future will be integrated seamlessly to design effective and efficient DNNs in the new edge intelligence era.”
From Bringing ai to edge: from deep learning’s perspective (2022) · cites this paper as [25]
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 15×
“In 2015, the ImageNet winning entry, ResNet [15], exceeded human-level accuracy with a top-5 error rate4 below 5%.”
InfluentialVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 10×
“Used in VGG-16. decompose 5x5 filter 5x1 filter 1x5 filter Apply sequentially (b) Constructing a 5×5 support from 1×5 and 5×1 filter.”
InfluentialImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 10×
“this, local response normalization (LRN) [3] was used, which”
BackgroundMethodInfluentialHochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 1×
“In contrast, recurrent neural networks (RNNs), of which Long Short-Term Memory networks (LSTMs) [38] are a popular variant, have internal memory to allow long-term dependencies to affect the output.”
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
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 4×
“Table 2 Summary of Popular DNNs [3], [15], [48], [50], [51]”
InfluentialGoing deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380 · cited 8×
“†ACCURACY IS MEASURED BASED ON TOP-5 ERROR ON IMAGENET [14].”
InfluentialFully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834
Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.81 · Cited by 32,684
Research works printed but not linked: 34
- 10.13140/RG.2.2.18893.74727 · DOI · Paper page
- Li, Stanford CS Class CS231n: Convolutional Neural Networks for Visual Recognition
- 2016, Complete Visual Networking Index (VNI) Forecast
- Woodhouse, 2016, Big, Big, Big Data: Higher and Higher Resolution Video Surveillance
- Maas, Rectifier nonlinearities improve neural network acoustic models, 1
- Caffe LeNet MNIST
- Caffe Model Zoo
- Matconvnet Pretrained Models
- TensorFlow-Slim Image Classification Library
- Deep Learning Frameworks
- Yann, The MNIST DATABASE of Handwritten Digits
- Krizhevsky, The CIFAR-10 Dataset
- Krizhevsky, 2010, Convolutional deep belief networks on cifar-10, Unpublished Manuscript, 40, 1
- Pascal VOC Data Sets
- Microsoft Common Objects in Context (COCO) Dataset
- Google Open Images
- YouTube-8M
- AudioSet
- 2017, Intel Architecture Instruction Set Extensions Programming Reference
- Condon, Facebook unveils Big Basin, new server geared for deep learning
- Intel Math Kernel Library
- 10.1145/1816038.1815993 · DOI
- 10.1145/2872887.2750389 · DOI
- 10.1145/2644865.2541967 · DOI · Paper page
- Standard, High bandwidth memory (HBM) DRAM
- 10.1145/3093336.3037702 · DOI
- Wilson, International technology roadmap for semiconductors (ITRS)
- 10.1109/icvd.2004.1260946 · doi:10.1109/icvd.2004.1260946
- Lee, A 2.5 GHz 7.7 TOPS/W switched-capacitor matrix multiplier with co-designed local memory in 40 nm, 418
- Lee, LogNet: Energy-efficient neural networks using logrithmic computations, 5900
- DNN Energy Estimation
- 10.1145/3140659.3080254 · DOI · Paper page
- Iandola, SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <1 MB model size
- Benchmarking DNN Processors
Cited by
Items 801–900 of 1,011
Chen, J., Lange, T., Andjelkovic, M. S., Simevski, A., Lu, L., & Krstic, M. D. (2022) · IEEE Transactions on Emerging Topics in Computing · doi:10.1109/tetc.2022.3147376 · Cited by 27
Dampfhoffer, M., Mesquida, T., Valentian, A., & Anghel, L. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-15934-3_30 · Cited by 30
Multiple-Mode-Supporting Floating-Point FMA Unit for Deep Learning Processors
Tan, H., Tong, G., Huang, L., Xiao, L., & Xiao, N. (2022) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2022.3226185 · Cited by 31
“However, multiply–add operations account for the majority of the calculation required in the deep learning algorithms, especially the convolution (CONV) and general matrix-multiply (GEMM) [12].”
BackgroundFang, C., Wang, C., Zhao, S., Tian, F., Yang, J., & Sawan, M. (2023) · IEEE Transactions on Biomedical Circuits and Systems · doi:10.1109/tbcas.2023.3279367 · Cited by 27
“IMPLEMENTING Artificial Neural Network (ANN) model on mobile devices with limited computation and storage resources has become possible in recent years, thanks to the development from both hardware [1] and software [2] sides.”
BackgroundData-Driven Control: Theory and Applications
Soudbakhsh, D., Annaswamy, A. M., Wang, Y., Brunton, S. L., Gaudio, J. E., Hussain, H. S., Vrabie, D., Drgoňa, J., & Filev, D. (2023) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc55779.2023.10156081 · Cited by 37
“While the approximation in (51) includes only one hidden layer, a deep neural network incorporates several additional layers, with the number of layers enabling approximations of more complex nonlinearities [54].”
BackgroundSeok, M., Kim, W., & Kim, J. (2023) · Healthcare · doi:10.3390/healthcare11212881 · Cited by 29
Benchmarking Delay and Energy of Neural Inference Circuits
Nikonov, D. E., & Young, I. A. (2019) · IEEE Journal on Exploratory Solid-State Computational Devices and Circuits · doi:10.1109/jxcdc.2019.2956112 · Cited by 32
“DIGITAL NEURAL ACCELERATORS There is another type of neural chip being fabricated, which is commonly called neural accelerators [3].”
BackgroundLearning Sparse Low-Precision Neural Networks With Learnable Regularization
Choi, Y., El‐Khamy, M., & Lee, J. (2020) · IEEE Access · doi:10.1109/access.2020.2996936 · Cited by 53
“In particular, we focus on the fixed-point implementation of convolutional and fully-connected layers, since they are the dominant parts of computational costs and memory requirements in DNNs (see [2, Table II]).”
BackgroundInfluentialRecent Advancements in Emerging Neuromorphic Device Technologies
Woo, J., Kim, J. H., Im, J., & Moon, S. E. (2020) · Advanced Intelligent Systems · doi:10.1002/aisy.202000111 · Cited by 37
Acceleration of Convolutional Neural Network Using FFT-Based Split ConvolutionsPreprint
Chitsaz, K., Hajabdollahi, M., Karimi, N., Samavi, S., & Shirani, S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.12621 · Cited by 31
“This problem is exacerbated in situations where there is a lack of available hardware resources [4].”
BackgroundImproved reservoir characterization of thin beds by advanced deep learning approach
Manzoor, U., Ehsan, M., Hussain, M., & Bashir, Y. (2024) · Applied Computing and Geosciences · doi:10.1016/j.acags.2024.100188 · Cited by 26
Optimum Selection of DNN Model and Framework for Edge Inference
Velasco-Montero, D., Fernández‐Berni, J., Carmona‐Galán, R., & Rodriguez-Vazquez, A. (2018) · IEEE Access · doi:10.1109/access.2018.2869929 · Cited by 25
“The interest of both industry and academia in DL is spurring the development of a myriad of software environments and hardware platforms tailored for DNNs [6].”
BackgroundInfluentialA Data-Center FPGA Acceleration Platform for Convolutional Neural Networks
Yu, X., Gao, J., Wang, Y., Miao, J., Wu, E., Zhang, H., Meng, Y., Zhang, B., Min, B., & Chen, D. (2019) · International Conference on Field Programmable Logic and Applications (FPL) · doi:10.1109/fpl.2019.00032 · Cited by 31
“Inspired by ResNet [5] and ShuffleNet [10], the HCNet begins with the convolution and pooling layers changing the input image from 224×224 into 56×56 with 32 channels.”
BackgroundMethodOptimized Fused Floating-Point Many-Term Dot-Product Hardware for Machine Learning Accelerators
Kaul, H., Anders, M. A., Mathew, S. K., Kim, S., & Krishnamurthy, R. K. (2019) · IEEE Symposium on Computer Arithmetic (ARITH) · doi:10.1109/arith.2019.00021 · Cited by 26
Clothing Classification Using Deep CNN Architecture Based on Transfer Learning
Elleuch, M., Mezghani, A., Khemakhem, M., & Kherallah, M. (2020) · Advances in intelligent systems and computing · doi:10.1007/978-3-030-49336-3_24 · Cited by 28
Scherer, M., Rutishauser, G., Cavigelli, L., & Benini, L. (2021) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2021.3075420 · Cited by 31
Sakatani, K., Oyama, K., & Hu, L. (2020) · Frontiers in Neurology · doi:10.3389/fneur.2020.588140 · Cited by 28
“CH 18 (7) 1 (0) 0 (0) 6 (5) 0 (1) 6 (1) 0 (1) 0 (0) 6 (0) 1 (0) 0 (0) 0 (0) 3 (1) 41 (16)”
BackgroundMethodOrganization of machine learning based product development as per ISO 26262 and ISO/PAS 21448
Radlak, K., Szczepankiewicz, M., Jones, T. J., & Serwa, P. (2020) · IEEE Pacific Rim International Symposium on Dependable Computing (PRDC) · doi:10.1109/prdc50213.2020.00022 · Cited by 30
“mance in many of the computer vision tasks involved in autonomous driving [33]–[35].”
BackgroundZaghari, N., Fathy, M., Jameii, S. M., Sabokrou, M., & Shahverdy, M. (2020) · The Journal of Supercomputing · doi:10.1007/s11227-020-03399-4 · Cited by 31
“It is proven that LSTMs can outperform standard feedforward neural networks on such tasks [16, 17].”
BackgroundEVEREST: A design environment for extreme-scale big data analytics on heterogeneous platforms
Pilato, C., Bohm, S., Brocheton, F., Castrillón, J., Cevasco, R., Cima, V., Cmar, R., Diamantopoulos, D., Ferrandi, F., Martinovič, J., et al. (2021) · Design, Automation, and Test in Europe · doi:10.23919/date51398.2021.9473940 · Cited by 31
Designing Efficient and High-Performance AI Accelerators With Customized STT-MRAM
Mishty, K., & Sadi, M. (2021) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2021.3105958 · Cited by 29
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
“While current models are generally trained and run on general-purpose digital processors such as CPUs and GPUs, the rapid growth in both size and scope of these networks has fostered novel hardware architectures aiming to optimize speed and energy-efficiency, specifically targeting either neural network training or in…”
BackgroundA bibliometric analysis of off-line handwritten document analysis literature (1990–2020)
Ruiz-Parrado, V., Heradio, R., Aranda‐Escolástico, E., Sánchez, Á., & Vélez, J. F. (2021) · Pattern Recognition · doi:10.1016/j.patcog.2021.108513 · Cited by 31
Omosebi, O. A., Oldenburg, C. M., & Reagan, M. T. (2022) · International journal of greenhouse gas control · doi:10.1016/j.ijggc.2021.103562 · Cited by 27
Designing efficient convolutional neural network structure: A survey
Mi, J., Feng, J., & Huang, K.-Y. (2022) · Neurocomputing · doi:10.1016/j.neucom.2021.08.158 · Cited by 35
A Codesigned Integrated Photonic Electronic Neuron
Marinis, L. D., Catania, A., Castoldi, P., Contestabile, G., Bruschi, P., Piotto, M., & Andriolli, N. (2022) · IEEE Journal of Quantum Electronics · doi:10.1109/jqe.2022.3177793 · Cited by 27
HW/SW Co-Design for Reliable TCAM- Based In-Memory Brain-Inspired Hyperdimensional Computing
Thomann, S., Genßler, P. R., & Amrouch, H. (2023) · IEEE Transactions on Computers · doi:10.1109/tc.2023.3248286 · Cited by 30
“tributes to the total energy consumption [1] and does form the key bottleneck.”
BackgroundGraphPrior: Mutation-based Test Input Prioritization for Graph Neural Networks
Dang, X., Li, Y., Papadakis, M., Klein, J., Bissyandé, T. F., & Traon, Y. L. (2023) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3607191 · Cited by 40
“(5) DNGP leverages a DNN model [78] as the ranking model.”
MethodBiswas, S. D., Kosta, A. K., Liyanagedera, C., Apolinario, M. P. E., & Roy, K. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00586 · Cited by 46
“autonomous navigation systems with significantly higher compute requirements and unlimited resources [3] still fail to replicate the comprehensive real-time scene understanding achieved by these tiny biological systems with extremely limited power budgets.”
BackgroundFine-grained energy profiling for deep convolutional neural networks on the Jetson TX1
Rodrigues, C. F., Riley, G. D., & Luján, M. (2017) · IEEE International Symposium on Workload Characterization · doi:10.1109/iiswc.2017.8167764 · Cited by 26
“• Many existing deep learning frameworks such as Caffe, Tensorflow, Torch, Theano and others allow users to define the software environment within which the inference application is executed [55].”
BackgroundA Mixed Signal Architecture for Convolutional Neural Networks
Lou, Q., Pan, C., McGuinness, J., Horváth, A., Naeemi, A., Niemier, M. T., & Hu, X. S. (2019) · ACM Journal on Emerging Technologies in Computing Systems · doi:10.1145/3304110 · Cited by 28
“A recent review paper summarized these approaches in [57].”
BackgroundA Quick Survey on Large Scale Distributed Deep Learning Systems
Zhang, Z., Yin, L., Peng, Y., & Li, D. (2018) · IEEE International Conference on Parallel and Distributed Systems (ICPADS) · doi:10.1109/padsw.2018.8644613 · Cited by 33
“Acceleration on single platform impacts on distributed systems significantly[14][15].”
BackgroundA Primer on Deep Learning Architectures and Applications in Speech Processing
Ogunfunmi, T., Ramachandran, R. P., Togneri, R. B., Zhao, Y., & Xia, X. (2019) · Circuits Systems and Signal Processing · doi:10.1007/s00034-019-01157-3 · Cited by 29
“Reference [88] gives a very comprehensive overview of the state of the art in this very important field.”
BackgroundInfluentialImplementation of DNNs on IoT devices
Zhang, Z., & Kouzani, A. Z. (2019) · Neural Computing and Applications · doi:10.1007/s00521-019-04550-w · Cited by 33
Development of CNN-based visual recognition air conditioner for smart buildings
Huang, Q., & Hao, K. (2020) · Journal of Information Technology in Construction · doi:10.36680/j.itcon.2020.021 · Cited by 27
“Then, various neural network architectures have been developed for many years (Sze et al., 2017; Gu et al., 2018; Zheng et al., 2020).”
BackgroundAnalysis of Edge-Optimized Deep Learning Classifiers for Radar-Based Gesture Recognition
Chmurski, M., Zubert, M., Bierzynski, K., & Santra, A. (2021) · IEEE Access · doi:10.1109/access.2021.3081353 · Cited by 27
Efficient Winograd Convolution via Integer ArithmeticPreprint
Meng, L., & Brothers, J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1901.01965 · Cited by 41
“Custom hardware (Sze et al., 2017; Han et al., 2016; Nvidia) with reduced-precision has also been designed and built for fast inference.”
BackgroundFrom Algorithm to Hardware: A Survey on Efficient and Safe Deployment of Deep Neural Networks
Geng, X., Wang, Z., Chen, C., Xu, Q., Xu, K., Jin, C., Gupta, M., Yang, X., Chen, Z., Sabry Aly, M. M., et al. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2024.3394494 · Cited by 26
“For example, there is a growing trend to combine KD with quantization and pruning.”
Chuang, Y.-C., Chiu, W., Chang, R. Y., & Lai, Y. (2023) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2023.3259109 · Cited by 25
“Because the neurons, the basic unit of neural networks, are computed by multiplications and additions, MACs are a common time efficiency metric for deep learning inference [49].”
BackgroundChen, G., Sun, M., Li, J., Wang, J., Shen, Y., Liang, D., & Xiao, R. (2024) · Energy · doi:10.1016/j.energy.2024.130966 · Cited by 25
Zhang, Y., Li, J., Lin, S., Zhao, J., Xiong, Y., & Wei, D.-Q. (2024) · Journal of Cheminformatics · doi:10.1186/s13321-024-00862-9 · Cited by 27
“…of machine learning (ML) and deep learning (DL) algorithms like Random Forest (RF) [4], Support Vector Machine (SVM) [5], Deep Neural Network (DNN) [6], Gradient Boosting Decision Tree (GBDT) [7] has speeded up the CPI identification process by enabling the development of novel compounds…”
BackgroundSemantic Image Encoding and Communication for Earth Observation With LEO Satellites
Bui, V.-P., Dinh, T. Q., Leyva‐Mayorga, I., Pandey, S. R., Lagunas, E., & Popovski, P. (2024) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2024.3451724 · Cited by 33
“Consequently, the total computational energy outlay is contingent solely upon the number of MACs [47].”
Accelerating Zero-Knowledge Proofs Through Hardware-Algorithm Co-Design
Samardzic, N., Langowski, S., Devadas, S., & Sánchez, D. (2024) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro61859.2024.00035 · Cited by 42
“For example, the CPU spends 2% of time on SpMV; this may seem low, but if we delegated SpMV to the CPU, this would limit speedups to 50 × , far below NoCap’s 586 × speedup.”
Channel Estimation in OTFS Systems by Leveraging Differential Modulation
Qing, C., Liu, Z., Ling, G., Hu, W., & Du, P. (2024) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2024.3522940 · Cited by 28
Alsharif, A. H., Wang, J., Isa, S. M., Salleh, N. Z. M., Dawas, H. A., & Alsharif, M. H. (2025) · Future Business Journal · doi:10.1186/s43093-025-00591-x · Cited by 50
“Moreover, the ability of DNNs to analyze large data-sets allows for identifying complex patterns in consumer behavior [323], which can inform marketing strategies.”
Boutros, A., Eldafrawy, M. M., Yazdanshenas, S., & Betz, V. (2019) · ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA) · doi:10.1145/3289602.3293912 · Cited by 29
“Recent CNN models consist of up to a thousand convolutional layers, followed by at least one fully-connected layer [22].”
MethodKim, J.-S., & Yang, J.-S. (2019) · ACM/IEEE Design Automation Conference (DAC) · doi:10.1145/3316781.3317805 · Cited by 26
On the Automatic Exploration of Weight Sharing for Deep Neural Network Compression
Dupuis, E., Novo, D., O’Connor, I., & Bosio, A. (2020) · Design, Automation, and Test in Europe · doi:10.23919/date48585.2020.9116350 · Cited by 27
“single step, but this technique lacks a clear representation of the data range, which is a crucial element to avoid accuracy losses [1].”
BackgroundInfluentialMcGrath, J. W., Neville, J., Stewart, T. I., & Cronin, J. B. (2020) · Proceedings of the Institution of Mechanical Engineers Part P Journal of Sports Engineering and Technology · doi:10.1177/1754337120959754 · Cited by 40
“DNNs have been shown to be capable of learning high-level features with more complexity and abstraction compared to NNs when using a large number of samples to train.(40) Further work is needed to investigate the effectiveness of deep learning methods, especially with higher sampling rate IMUs which produce more volum…”
BackgroundLow-Cost Error Detection in Deep Neural Network Accelerators with Linear Algorithmic Checksums
Ozen, E., & Orailoğlu, A. (2020) · Journal of Electronic Testing · doi:10.1007/s10836-020-05920-2 · Cited by 27
Huang, C., Liu, P., & Fang, L. (2021) · Applied Intelligence · doi:10.1007/s10489-020-02109-0 · Cited by 30
Hardware design and the competency awareness of a neural network
Ding, Y., Jiang, W., Lou, Q., Liu, J., Xiong, J., Hu, X. S., Xu, X., & Shi, Y. (2020) · Nature Electronics · doi:10.1038/s41928-020-00476-7 · Cited by 24
“With the uncertainty estimation quality as the key factor, such an early exit method boosts the throughput with neglectable performance loss 68 .”
BackgroundMethodKilcioglu, E., Mirghasemi, H., Stupia, I., & Vandendorpe, L. (2021) · IEEE Access · doi:10.1109/access.2021.3084689 · Cited by 27
“The convolution operation in convolutional layers can be decomposed into lots of multiply-accumulate (MAC) operations [44], [45].”
BackgroundTwo Sides of the Same Coin: Boons and Banes of Machine Learning in Hardware Security
Liu, W., Chang, C.-H., Wang, X., Liu, C., Fung, J. M., Ebrahimabadi, M., Karimi, N., Meng, X., & Basu, K. (2021) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2021.3084400 · Cited by 34
“For instance, Open source Basic Linear Algebra Subprograms (OpenBLAS) and NVIDIA CUDA® DNN library cuDNN are parallelized backends for GEMM execution on CPU and GPU, respectively [51].”
A Federated Learning Approach to Frequent Itemset Mining in Cyber-Physical Systems
Ahmed, U., Srivastava, G., & Lin, J. C. (2021) · Journal of Network and Systems Management · doi:10.1007/s10922-021-09609-5 · Cited by 30
A defense method against backdoor attacks on neural networks
Kaviani, S., Shamshiri, S., & Sohn, I. (2022) · Expert Systems with Applications · doi:10.1016/j.eswa.2022.118990 · Cited by 27
Mohammadifar, A., Gholami, H., & Golzari, S. (2022) · Environmental Science and Pollution Research · doi:10.1007/s11356-022-24065-7 · Cited by 28
“DL models apply an unsupervised pre-training and a supervised fine-tuning strategy to learn hierarchical features (Sze et al. 2017).”
MethodAlqadhi, S. D., Mallick, J., Alkahtani, M. Q., Ahmad, I., Alqahtani, D. A., & Hằng, H. T. (2023) · Natural Hazards · doi:10.1007/s11069-023-06357-4 · Cited by 37
“Deep neural networks (DNNs) are powerful artificial neural networks characterized by their deep architecture with multiple hidden layers situated between the input and output layers (Sze et al. 2017; Montavon et al. 2018).”
Spiking-Fer: Spiking Neural Network for Facial Expression Recognition With Event Cameras
Barchid, S., Allaert, B., Aissaoui, A., Mennesson, J., & Djéraba, C. (2023) · International Workshop on Content-Based Multimedia Indexing (CBMI) · doi:10.1145/3617233.3617235 · Cited by 26
“MAC operation in ANNs requires one addition (32bit FP ADD) and one FP multiplication (32bit FP MULT) [35], whereas SNNs require only one FP addition per MAC operation Table 3 reports the mean inference energy estimation for each dataset.”
MethodA deep multi-task learning approach for ECG data analysis
Ji, J., Chen, X., Luo, C., & Li, P. (2018) · IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI ...) · doi:10.1109/bhi.2018.8333385 · Cited by 26
“Since deep learning based methods have strong capabilities in nonlinear functional learning, they have been widely used in multi-dimensional signal processing problems like computer vision, speech recognition, and natural language processing [5].”
BackgroundEnabling Energy-Efficient Unsupervised Monocular Depth Estimation on ARMv7-Based Platforms
Peluso, V., Cipolletta, A., Calimera, A., Poggi, M., Tosi, F., & Mattoccia, S. (2019) · Design, Automation, and Test in Europe · doi:10.23919/date.2019.8714893 · Cited by 32
“The actual benefit lies in the ability to store the same amount of information with fewer bits which in turn enables a more efficient use of the memory bandwidth [21].”
BackgroundMemory-Reduced Network Stacking for Edge-Level CNN Architecture With Structured Weight Pruning
Moon, S., Byun, Y., Park, J., Lee, S., & Lee, Y. (2019) · IEEE Journal on Emerging and Selected Topics in Circuits and Systems · doi:10.1109/jetcas.2019.2952137 · Cited by 27
“However, when considering the tight energy constraints of edge-level devices, it is necessary to use a more efficient method for storing multiple CNN architectures [9].”
MethodGomatheeshwari, B., & Selvakumar, J. (2020) · Microprocessors and Microsystems · doi:10.1016/j.micpro.2020.102996 · Cited by 25
Analyzing the impact of soft errors in VGG networks implemented on GPUs
Wei, J., Ibrahim, Y., Qian, S., Wang, H., Liu, G., Yu, Q., Qian, R., & Shi, J. (2020) · Microelectronics Reliability · doi:10.1016/j.microrel.2020.113648 · Cited by 29
Speeding up Convolutional Neural Networks By Exploiting the Sparsity of Rectifier UnitsPreprint
Shi, S., & Chu, X. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1704.07724 · Cited by 34
“In order to exploit the advantage of SIMD techniques of processors (CPUs or GPUs), the calculation of convolutional layer is often mapped to matrix multiplication [5, 23].”
MethodLearning Low Precision Deep Neural Networks through Regularization.Preprint
Choi, Y., El‐Khamy, M., & Lee, J. (2018) · arXiv (Cornell University) · Cited by 29
Recent Developments in Low-Power AI Accelerators: A Survey
Åleskog, C., Grahn, H., & Borg, A. (2022) · Algorithms · doi:10.3390/a15110419 · Cited by 36
“For a more detailed explanation of the mentioned strategies and how DNNs are accelerated in hardware, we recommend the paper by Sze et al. [10].”
MethodDistributed digital twins for health monitoring: resource constrained aero-engine fleet management
Hartwell, A., Montana, F. J., Jacobs, W. R., Kadirkamanathan, V., Ameri, N., & Mills, A. R. (2024) · The Aeronautical Journal · doi:10.1017/aer.2024.23 · Cited by 25
Gligorić, Z., Görçün, Ö. F., Gligorić, M., Pamucar, D. S., Šimić, V., & Küçükönder, H. (2024) · Journal of King Saud University - Computer and Information Sciences · doi:10.1016/j.jksuci.2024.102079 · Cited by 31
Xia, Q., Wang, Q., Zou, Y., Chi, Y., Yan, Z., Meng, Q., Zhou, N., & Guerrero, J. M. (2024) · Journal of Energy Storage · doi:10.1016/j.est.2024.113477 · Cited by 25
Energy-Efficient and Accuracy-Aware DNN Inference With IoT Device-Edge Collaboration
Jiang, W., Han, H., Feng, D., Qian, L., Wang, Q., & Xia, X. (2025) · IEEE Transactions on Services Computing · doi:10.1109/tsc.2025.3536311 · Cited by 33
Embedded Artificial Intelligence: A Comprehensive Literature Review
Huang, X., Wang, H., Qin, S., & Tang, S.-K. (2025) · Electronics · doi:10.3390/electronics14173468 · Cited by 38
“ASICs are specialized chips meticulously designed for particular applications, offering the highest performance and lowest power consumption for their designated tasks [27,56].”
InfluentialMachine Learning Models of the Geospatial Distribution of Groundwater Quality: A Systematic Review
Mehrabi, M., Polya, D. A., & Han, Y. (2025) · Water · doi:10.3390/w17192861 · Cited by 27
Pruning Blocks for CNN Compression and Acceleration via Online Ensemble Distillation
Wang, Z., Lin, S., Xie, J., & Lin, Y. (2019) · IEEE Access · doi:10.1109/access.2019.2957203 · Cited by 24
“For more details, the reader is referred to a survey [71].”
BackgroundRobust and structural sparsity auto-encoder with L21-norm minimization
Li, R., Wang, X., Wen, Q., Song, Y., & Lei, L. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.02.051 · Cited by 24
Jiang, S., Ma, Z., Zeng, X., Xu, C., Zhang, M., Zhang, C., & Liu, Y. (2020) · IEEE Conference on Computer Communications (INFOCOM) · doi:10.1109/infocom41043.2020.9155435 · Cited by 32
“However, the model complexities pay for high computational resources [8], [9], leading to high end-to-end latency in CPU-based implementation.”
BackgroundPreVIous: A Methodology for Prediction of Visual Inference Performance on IoT Devices
Velasco Montero, D., Fernández Berni, J., Carmona Galán, R., & Rodríguez Vázquez, Á. B. (2020) · LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) · Cited by 24
Chang, X., Li, Y., Oymak, S., & Thrampoulidis, C. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i8.16859 · Cited by 24
New universal sustainability metrics to assess edge intelligence
Lenherr, N., Pawlitzek, R., & Michel, B. (2021) · Sustainable Computing: Informatics and Systems · doi:10.1016/j.suscom.2021.100580 · Cited by 23
Performance Analysis of Deep Neural Networks Using Computer Vision
Sindhwani, N., Anand, R., Meivel, S., Shukla, R., Yadav, M. P., & Yadav, V. (2021) · EAI Endorsed Transactions on Industrial Networks and Intelligent Systems · doi:10.4108/eai.13-10-2021.171318 · Cited by 27
“As defined in [35], these networks are made up of many layers and consist of an input layer, at least one hidden layer and an output layer.”
BackgroundPower Management for Connected EVs Using a Fuzzy Logic Controller and Artificial Neural Network
Angundjaja, C. Y., Wang, Y., & Jiang, W. (2021) · Applied Sciences · doi:10.3390/app12010052 · Cited by 24
“If the SC will still supply most of the power.”
Akhtar, M. B. (2022) · HighTech and Innovation Journal · doi:10.28991/hij-2022-03-01-01 · Cited by 25
“CNN [34, 35] has achieved outstanding results in image classification and detection tasks [36, 37].”
BackgroundHalide perovskite based synaptic devices for neuromorphic systems
Beom, K., Fan, Z., Li, D., & Newman, N. (2022) · Materials Today Physics · doi:10.1016/j.mtphys.2022.100667 · Cited by 29
Comparing deep learning methods to predict the remaining useful life of lithium-ion batteries
Zraibi, B., Mansouri, M., & Loukili, S. E. (2022) · Materials Today Proceedings · doi:10.1016/j.matpr.2022.04.082 · Cited by 27
Jiang, H., Wang, H., Su, Y., Kang, Q., Meng, X., Yan, L., & Ma, T. (2022) · Journal of Power Sources · doi:10.1016/j.jpowsour.2022.231818 · Cited by 24
Review and application of Edge AI solutions for mobile collaborative robotic platforms
Ramasubramanian, A. K., Mathew, R., Preet, I., & Papakostas, N. (2022) · Procedia CIRP · doi:10.1016/j.procir.2022.05.112 · Cited by 28
Hardware Accelerators for Real-Time Face Recognition: A Survey
Baobaid, A., Meribout, M., Tiwari, V. K., & Pena, J. P. (2022) · IEEE Access · doi:10.1109/access.2022.3194915 · Cited by 32
Putra, R. V. W., Hanif, M. A., & Shafique, M. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2022.937782 · Cited by 29
Yan, Z., Hu, X. S., & Shi, Y. (2022) · International Conference on Computer Aided Design · doi:10.1145/3508352.3549360 · Cited by 26
“3549360 can reduce data movement by performing in-situ weight data access [29].”
BackgroundCA-SpaceNet: Counterfactual analysis for 6D pose estimation in space
Wang, S., Wang, S., Jiao, B., Yang, D., Su, L., Zhai, P., Chen, C., & Zhang, L. (2022) · 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981172 · Cited by 24
“DNNs run with low precision operations during inference provide power and memory advantages over full precision, and it also benefits low-bit-width artificial intelligence chip design [37], [38].”
BackgroundMalaria Disease Cell Classification With Highlighting Small Infected Regions
Yebasse, M. T., Cheoi, K. J., & Ko, J. (2023) · IEEE Access · doi:10.1109/access.2023.3245025 · Cited by 24
“Several meth-ods have recently been introduced to enable a faster and more stable training in deep learning [26], [27], [28].”
BackgroundDependable DNN Accelerator for Safety-Critical Systems: A Review on the Aging Perspective
Moghaddasi, I., Gorgin, S., & Lee, J. (2023) · IEEE Access · doi:10.1109/access.2023.3300376 · Cited by 30
“DNN comprises multiple types of layers, including convolutional, pooling, normalization, and fully-connected sublayers [30].”
BackgroundLiu, J., Zhang, T., & Sun, S. (2023) · Geoscience Frontiers · doi:10.1016/j.gsf.2023.101735 · Cited by 27
Hardware-Efficient Logarithmic Floating-Point Multipliers for Error-Tolerant Applications
Niu, Z., Zhang, T., Jiang, H., Cockburn, B. F., Liu, L., & Han, J. (2023) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2023.3326329 · Cited by 29
“The substantial increase in power becomes a limiting factor for computation-intensive applications, such as machine learning (ML) and digital signal processing (DSP), in resource-constrained devices [2], [3].”
BackgroundLi, W., Hu, A., Xu, N., & He, G. (2024) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2024.3350661 · Cited by 38
“Note that the matrix-matrix multiplications required by prompting stage can also be decomposed into MVMs. Ingeneral, fixed-point (integer) arithmetic units have lower area cost and power consumption than their floating-point counterparts [48], [49].”
BackgroundAdaptive and Communication-Efficient Zeroth-Order Optimization for Distributed Internet of Things
Dang, Q., Yang, S., Liu, Q., & Ruan, J. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3441691 · Cited by 28
Low Resolution Image Fish Classification Using Convolutional Neural Network
Rachmatullah, M. N., & Supriana, I. (2018) · International Conference on Advanced Informatics: Concepts, Theory and Applications (ICAICTA) · doi:10.1109/icaicta.2018.8541313 · Cited by 24
“The feature learning result from CNN is a feature map (fmap), which holds important and unique information from the data [6].”
BackgroundThe Facial Stress Recognition Based on Multi-histogram Features and Convolutional Neural Network
Prasetio, B. H., Tamura, H., & Tanno, K. (2018) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc.2018.00157 · Cited by 24
“The DNN is more accurate and efficient than typically neural networks [16].”
MethodAn Efficient FPGA-based Overlay Inference Architecture for Fully Connected DNNs
Abdelsalam, A. M., Boulet, F., Demers, G., Langlois, J. M. P., & Farida Cheriet, A. (2018) · International Conference on ReConFigurable Computing and FPGAs (ReConFig) · doi:10.1109/reconfig.2018.8641735 · Cited by 25
“These techniques have been combined in several DNN hardware accelerators on FPGAs [5], [6], [8].”
BackgroundMethodLin, Y., Lee, F.-M., Lee, M.-H., Chen, W.-C., Lung, H.-L., Wang, K.-C., & Lu, C.-Y. (2018) · Technical digest - International Electron Devices Meeting/Technical digest · doi:10.1109/iedm.2018.8614688 · Cited by 24
Metadata
- Authors
- Sze, V., Chen, Y.-H., Yang, T.-J., & Emer, J. S.
- Affiliations
- Massachusetts Institute of Technology (US)
- Nvidia (United States) (US)
- Venue
- Proceedings of the IEEE
- Published
- 2017-12
- Volume
- 105
- Issue
- 12
- Pages
- 2295-2329
- ISSN
- 0018-9219 (ISSN-L)
- 1558-2256
- OpenAlex
- W2604319603
- Cited by
- 4,215
- References
- 157
- Citations by year
- 2026: 504 · 2025: 523 · 2024: 474 · 2023: 466 · 2022: 508 · 2021: 568 · 2020: 551 · 2019: 392 · 2018: 199 · 2017: 28
- FWCI
- 132.75
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Proceedings of the IEEE (publishedVersion)
- Funders
- DARPA YFA
- MIT CICS
- gift from Nvidia
- gift from Intel
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Memory and Neural Computing
- Keywords
- deep neural network, energy efficiency, throughput, hardware accelerators, computation cost reduction, benchmarking metrics
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.