Deep Neural Network Approximation for Custom Hardware: Where We've Been, Where We're Going
Open accessThe DOI matches its Crossref record; 39 of the 39 references it cites were found in a registry or an index.
Abstract
Deep neural networks have proven to be particularly effective in visual and audio recognition tasks. Existing models tend to be computationally expensive and memory intensive, however, and so methods for hardware-oriented approximation have become a hot topic. Research has shown that custom hardware-based neural network accelerators can surpass their general-purpose processor equivalents in terms of both throughput and energy efficiency. Application-tailored accelerators, when co-designed with approximation-based network training methods, transform large, dense, and computationally expensive networks into small, sparse, and hardware-efficient alternatives, increasing the feasibility of network deployment. In this article, we provide a comprehensive evaluation of approximation methods for high-performance network inference along with in-depth discussion of their effectiveness for custom hardware implementation. We also include proposals for future research based on a thorough analysis of current trends. This article represents the first survey providing detailed comparisons of custom hardware accelerators featuring approximation for both convolutional and recurrent neural networks, through which we hope to inspire exciting new developments in the field.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., & Adam, H. (2017) · arXiv · doi:10.48550/arxiv.1704.04861 · Cited by 19,015
Distilling the Knowledge in a Neural Network
Hinton, G., Vinyals, O., & Dean, J. (2015) · arXiv · doi:10.48550/arxiv.1503.02531 · Cited by 14,204
Large-Scale Video Classification with Convolutional Neural Networks
Karpathy, A., Toderici, G., Shetty, S., Leung, T., Sukthankar, R., & Fei-Fei, L. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.223 · Cited by 6,354
Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi, A. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.11231 · Cited by 4,636 · cited 1×
“ey have now been employed in a vast range of application domains, from image classication [137] and object detection [90] to autonomous driving [19] and drone navigation [42].”
MethodIn-Datacenter Performance Analysis of a Tensor Processing Unit
Jouppi, N. P., Young, C., Patil, N., Patterson, D., Agrawal, G., Bajwa, R., Bates, S., Bhatia, S., Boden, N., Borchers, A., et al. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080246 · Cited by 4,537 · cited 5×
“Their timeconsuming and resource-demanding design and fabrication processes, however, make it hard for them to keep up with the rapid evolution of DNN algorithms [28, 66].”
BackgroundMethodInfluentialXNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks
Rastegari, M., Ordonez, V., Redmon, J., & Farhadi, A. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_32 · Cited by 4,408
Efficient Processing of Deep Neural Networks: A Tutorial and Survey
Sze, V., Chen, Y.-H., Yang, T.-J., & Emer, J. S. (2017) · Proceedings of the IEEE · doi:10.1109/jproc.2017.2761740 · Cited by 4,215 · cited 2×
“For example, a small LeNet-5 model targeing the simple MNIST handwrien digitclassication task requires 680 kop/cl (thousand arithmetic operations per classication, where an arithmetic operation is either an addition or multiplication), while a VGG16 implementation executing the 1000-class ImageNet task requires 31…”
MethodQuantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference
Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., & Kalenichenko, D. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00286 · Cited by 4,114
Eyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural Networks
Chen, Y.-H., Krishna, T., Emer, J. S., & Sze, V. (2017) · IEEE Journal of Solid-State Circuits · doi:10.1109/jssc.2016.2616357 · Cited by 3,247 · cited 6×
“Eyeriss, EIE and Laconic’s authors achieved benefits from pruning using similar strategies to those employed by Cnvlutin’s.”
BackgroundMethodInfluentialProduct Quantization for Nearest Neighbor Search
Jégou, H., Douze, M., & Schmid, C. (2011) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2010.57 · Cited by 2,956 · cited 3×
“The authors of PQ-CNN presented a hardware-software framework for compressing and accelerating CNNs on FPGAs using product quantisation [64], adopting a similar idea to that used in Quantised CNN [149, 156].”
MethodPruning Filters for Efficient ConvNets
Li, H., Kadav, A., Durdanovic, I., Samet, H., & Graf, H. P. (2016) · arXiv · doi:10.48550/arxiv.1608.08710 · Cited by 2,928
FitNets: Hints for Thin Deep Nets
Romero, A., Ballas, N., Kahou, S. E., Chassang, A., Gatta, C., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1412.6550 · Cited by 2,859
Learning Efficient Convolutional Networks through Network Slimming
Liu, Z., Li, J., Shen, Z., Huang, G., Yan, S., & Zhang, C. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.298 · Cited by 2,761
Channel Pruning for Accelerating Very Deep Neural Networks
He, Y., Zhang, X., & Sun, J. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.155 · Cited by 2,647 · cited 1×
“[57] and Su et al.”
Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
Bengio, Y., Léonard, N., & Courville, A. (2013) · arXiv · doi:10.48550/arxiv.1308.3432 · Cited by 1,910
DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients
Zhou, S., Wu, Y., Ni, Z., Zhou, X., Wen, H., & Zou, Y. (2016) · arXiv · doi:10.48550/arxiv.1606.06160 · Cited by 1,690
Learning both Weights and Connections for Efficient Neural Networks
Han, S., Pool, J., Tran, J., & Dally, W. J. (2015) · arXiv · doi:10.48550/arxiv.1506.02626 · Cited by 1,481 · cited 7×
“Among them, ne-grained methods (DNS [50] and Network Pruning [55]) show more promising tradeos than coarse-grained alternatives (Structured Brain Damage [74] and Less is More [160]).”
BackgroundMethodInfluentialExploiting Linear Structure Within Convolutional Networks for Efficient Evaluation
Denton, R., Zaremba, W., Bruna, J., LeCun, Y., & Fergus, R. (2014) · arXiv · doi:10.48550/arxiv.1404.0736 · Cited by 1,414
Deep Learning with Limited Numerical Precision
Gupta, S., Agrawal, A., Gopalakrishnan, K., & Narayanan, P. (2015) · arXiv · doi:10.48550/arxiv.1502.02551 · Cited by 1,364 · cited 1×
“trained CNNs with 16-bit xed-point weight representation using stochastic rounding, achieving lossless compression for the MNIST and CIFAR-10 datasets [51].”
MethodFINN: A Framework for Fast, Scalable Binarized Neural Network Inference
Umuroglu, Y., Fraser, N. J., Gambardella, G., Blott, M., Leong, P., Jahre, M., & Vissers, K. (2017) · Proceedings of the 2017 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays · doi:10.1145/3020078.3021744 · Cited by 1,189
Speeding up Convolutional Neural Networks with Low Rank Expansions
Jaderberg, M., Vedaldi, A., & Zisserman, A. (2014) · Proceedings of the British Machine Vision Conference 2014 · doi:10.5244/c.28.88 · Cited by 1,164
EIE: Efficient Inference Engine on Compressed Deep Neural Network
Han, S., Liu, X., Mao, H., Pu, J., Pedram, A., Horowitz, M. A., & Dally, W. J. (2016) · 2016 ACM/IEEE 43rd Annual International Symposium on Computer Architecture (ISCA) · doi:10.1109/isca.2016.30 · Cited by 1,077
Compressing Neural Networks with the Hashing Trick
Chen, W., Wilson, J. T., Tyree, S., Weinberger, K. Q., & Chen, Y. (2015) · arXiv · doi:10.48550/arxiv.1504.04788 · Cited by 974
Cheng, Y., Wang, D., Zhou, P., & Zhang, T. (2018) · IEEE Signal Processing Magazine · doi:10.1109/msp.2017.2765695 · Cited by 817
Li, F., Liu, B., Wang, X., Zhang, B., & Yan, J. (2016) · arXiv · doi:10.48550/arxiv.1605.04711 · Cited by 732
Designing Energy-Efficient Convolutional Neural Networks Using Energy-Aware Pruning
Yang, T.-J., Chen, Y.-H., & Sze, V. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.643 · Cited by 732 · cited 1×
“7× with ImageNet on AlexNet against an Nvidia Titan X GPU equivalent [152].”
MethodFixed Point Quantization of Deep Convolutional Networks
Lin, D. D., Talathi, S. S., & Annapureddy, V. S. (2015) · arXiv · doi:10.48550/arxiv.1511.06393 · Cited by 684 · cited 1×
“[84] formulated and solved an SQNR-based optimisation problem to identify the optimal fixed-point precision per layer of a custom-designed CNN, showing that the proposed scheme offered over 1.”
MethodESE: Efficient Speech Recognition Engine with Sparse LSTM on FPGA
Han, S., Kang, J., Mao, H., Hu, Y., Li, X., Li, Y., Xie, D., Luo, H., Yao, S., Wang, Y., et al. (2017) · Proceedings of the 2017 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays · doi:10.1145/3020078.3021745 · Cited by 591 · cited 14×
“fpgaConvNet [142], Angel-eye [48], ESE [52] and works by Chang et al.”
BackgroundMethodInfluentialNetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications
Yang, T.-J., Howard, A., Chen, B., Zhang, X., Go, A., Sandler, M., Sze, V., & Adam, H. (2018) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-01249-6_18 · Cited by 572
Deep Learning with Low Precision by Half-Wave Gaussian Quantization
Cai, Z., He, X., Sun, J., & Vasconcelos, N. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.574 · Cited by 498 · cited 1×
“Going further, HWGQ addressed the problem of mismatching gradients between the binarised forward activation function, sign, and the backward activation function, hard tanh [16].”
BackgroundCan FPGAs Beat GPUs in Accelerating Next-Generation Deep Neural Networks?
Nurvitadhi, E., Venkatesh, G., Sim, J., Marr, D., Huang, R., Ong Gee Hock, J., Liew, Y. T., Srivatsan, K., Moss, D., Subhaschandra, S., et al. (2017) · Proceedings of the 2017 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays · doi:10.1145/3020078.3021740 · Cited by 467
Predicting Parameters in Deep Learning
Shakibi, B. (2014) · University of British Columbia · doi:10.14288/1.0165555 · Cited by 462
Towards Accurate Binary Convolutional Neural Network
Lin, X., Zhao, C., & Pan, W. (2017) · arXiv · doi:10.48550/arxiv.1711.11294 · Cited by 453 · cited 2×
“e authors of ReBNet implemented “residual binarisation” on FPGAs [44]: similar to ABC-NET’s aforementioned K-binarisation scheme [86].”
MethodNeuFlow: A runtime reconfigurable dataflow processor for vision
Farabet, C., Martini, B., Corda, B., Akselrod, P., Culurciello, E., & LeCun, Y. (2011) · CVPR 2011 WORKSHOPS · doi:10.1109/cvprw.2011.5981829 · Cited by 369
[DL] A Survey of FPGA-based Neural Network Inference Accelerators
Guo, K., Zeng, S., Yu, J., Wang, Y., & Yang, H. (2019) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/3289185 · Cited by 292
How to Train a Compact Binary Neural Network with High Accuracy?
Tang, W., Hua, G., & Wang, L. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.10862 · Cited by 241
WRPN: Wide Reduced-Precision Networks
Mishra, A., Nurvitadhi, E., Cook, J. J., & Marr, D. (2017) · arXiv · doi:10.48550/arxiv.1709.01134 · Cited by 227 · cited 1×
“y, a low-precision network can eventually match or even surpass the accuracy of its baseline. Further research is required to identify models that are particularly amenable to low-precision inference [96]. 4.2.2 Hardware Implementation. For inference, binary networks have several properties that enable elegant mapping…”
BackgroundInfluentialYang, Z., Moczulski, M., Denil, M., De Freitas, N., Song, L., & Wang, Z. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.173 · Cited by 215 · cited 2×
“nst an FP32 baseline [37]. An alternative transformation, the Adaptive Fastfood transform (AFT), achieved a compression ratio of 3.7, but only about 0.1 pp of accuracy loss with ImageNet, for AlexNet [154]. In an AFT, a weight matrixW is approximated as W AFT = SHGΠHB; in which S, G and B are trainable diagonal matric…”
MethodInfluentialDeep Convolutional Neural Network Inference with Floating-point Weights and Fixed-point Activations
Lai, L., Suda, N., & Chandra, V. (2017) · arXiv · doi:10.48550/arxiv.1703.03073 · Cited by 81 · cited 3×
“that it is weights’ representation range, rather than precision, that is crucial to the preservation of accuracy [72].”
BackgroundMethodInfluential
Cited by
Efficient Acceleration of Deep Learning Inference on Resource-Constrained Edge Devices: A Review
Shuvo, M. M. H., Islam, S. K., Cheng, J., & Morshed, B. I. (2023) · Proceedings of the IEEE · doi:10.1109/jproc.2022.3226481 · Cited by 436
Furano, G., Meoni, G., Dunne, A., Moloney, D., Ferlet-Cavrois, V., Tavoularis, A., Byrne, J., Buckley, L., Psarakis, M., Voss, K.-O., et al. (2020) · IEEE Aerospace and Electronic Systems Magazine · doi:10.1109/maes.2020.3008468 · Cited by 259
AI and ML Accelerator Survey and Trends
Reuther, A., Michaleas, P., Jones, M., Gadepally, V., Samsi, S., & Kepner, J. (2022) · 2022 IEEE High Performance Extreme Computing Conference (HPEC) · doi:10.1109/hpec55821.2022.9926331 · Cited by 109
Mokey: enabling narrow fixed-point inference for out-of-the-box floating-point transformer models
Zadeh, A. H., Mahmoud, M., Abdelhadi, A., & Moshovos, A. (2022) · Proceedings of the 49th Annual International Symposium on Computer Architecture · doi:10.1145/3470496.3527438 · Cited by 37
GOBO: Quantizing Attention-Based NLP Models for Low Latency and Energy Efficient Inference
Zadeh, A. H., Edo, I., Awad, O. M., & Moshovos, A. (2020) · 2020 53rd Annual IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1109/micro50266.2020.00071 · Cited by 10
Metadata
- Authors
- Wang, E., Davis, J. J., Zhao, R., Ng, H.-C., Niu, X., Luk, W., Cheung, P. Y. K., & Constantinides, G. A.
- Venue
- ACM Computing Surveys
- Published
- 2019-05-30
- Volume
- 52
- Issue
- 2
- Pages
- 1-39
- ISSN
- 0360-0300 (ISSN-L)
- 1557-7341
- OpenAlex
- W3123290820
- Cited by
- 166
- References
- 39
- FWCI
- 12.74
- Type
- journal article
- Language
- en
- Open access
- green
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.