DOI: 10.1109/cvpr.2017.574. Deep Learning with Low Precision by Half-Wave Gaussian Quantization

Deep Learning with Low Precision by Half-Wave Gaussian Quantization

10.1109/cvpr.2017.574

Crossref proceedings-article

IEEE

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (263)

Bibliography

Cai, Z., He, X., Sun, J., & Vasconcelos, N. (2017). Deep Learning with Low Precision by Half-Wave Gaussian Quantization. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 5406â5414.

Authors 4

Zhaowei Cai (first)
Xiaodong He (additional)
Jian Sun (additional)
Nuno Vasconcelos (additional)

References 37 Referenced 314

{'key': 'ref33', 'volume': 'abs 1409 1556', 'author': 'simonyan', 'year': '2014', 'journal-title': 'Very Deep Convolutional Networks for Large-scale Image Recognition'} / Very Deep Convolutional Networks for Large-scale Image Recognition by simonyan (2014)
10.1007/s11263-015-0816-y
{'key': 'ref31', 'first-page': '91', 'article-title': 'Faster R-CNN: towards real-time object detection with region proposal networks', 'author': 'ren', 'year': '2015', 'journal-title': 'NIPS'} / NIPS / Faster R-CNN: towards real-time object detection with region proposal networks by ren (2015)
{'key': 'ref30', 'first-page': '525', 'article-title': 'Xnornet: Imagenet classification using binary convolutional neural networks', 'author': 'rastegari', 'year': '2016', 'journal-title': 'ECCV'} / ECCV / Xnornet: Imagenet classification using binary convolutional neural networks by rastegari (2016)
{'key': 'ref37', 'volume': 'abs 1606 6160', 'author': 'zhou', 'year': '2016', 'journal-title': 'DoReFa-Net Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients'} / DoReFa-Net Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients by zhou (2016)
10.1109/ICASSP.2013.6638312
{'key': 'ref35', 'article-title': 'Improving the speed of neural networks on cpus', 'author': 'vanhoucke', 'year': '2011', 'journal-title': 'Proc Deep Learning and Unsupervised Feature Learning Workshop NIPS 2011'} / Proc Deep Learning and Unsupervised Feature Learning Workshop NIPS 2011 / Improving the speed of neural networks on cpus by vanhoucke (2011)
10.1109/CVPR.2015.7298594
{'key': 'ref10', 'first-page': '1135', 'article-title': 'Learning both weights and connections for efficient neural network', 'author': 'han', 'year': '2015', 'journal-title': 'NIPS'} / NIPS / Learning both weights and connections for efficient neural network by han (2015)
10.1109/ICCV.2015.123
10.1109/CVPR.2016.90
{'key': 'ref13', 'volume': 'abs 1207 580', 'author': 'hinton', 'year': '2012', 'journal-title': 'Improving Neural Networks by Preventing Co-adaptation of Feature Detectors'} / Improving Neural Networks by Preventing Co-adaptation of Feature Detectors by hinton (2012)
10.1214/aoms/1177703732
10.1016/S0893-6080(00)00026-5
{'key': 'ref16', 'volume': 'abs 1602 7360', 'author': 'iandola', 'year': '2016', 'journal-title': 'SqueezeNet AlexNet-level accuracy with 50x fewer parameters and < 1MB model size'} / SqueezeNet AlexNet-level accuracy with 50x fewer parameters and < 1MB model size by iandola (2016)
{'key': 'ref17', 'first-page': '448', 'article-title': 'Batch normalization: Accelerating deep network training by reducing internal covariate shift', 'author': 'ioffe', 'year': '2015', 'journal-title': 'ICML'} / ICML / Batch normalization: Accelerating deep network training by reducing internal covariate shift by ioffe (2015)
{'key': 'ref18', 'article-title': 'Speeding up convolutional neural networks with low rank expansions', 'author': 'jaderberg', 'year': '2014', 'journal-title': 'BMVC'} / BMVC / Speeding up convolutional neural networks with low rank expansions by jaderberg (2014)
10.1145/2647868.2654889
{'key': 'ref28', 'first-page': '807', 'article-title': 'Rectified linear units improve restricted boltzmann machines', 'author': 'nair', 'year': '2010', 'journal-title': 'ICML'} / ICML / Rectified linear units improve restricted boltzmann machines by nair (2010)
{'key': 'ref4', 'volume': 'abs 1602 2830', 'author': 'courbariaux', 'year': '2016', 'journal-title': 'Binarized neural networks Training neural networks with weights and activations constrained to +1 or ?1'} / Binarized neural networks Training neural networks with weights and activations constrained to +1 or ?1 by courbariaux (2016)
10.1109/TIT.1982.1056489
{'key': 'ref3', 'first-page': '3123', 'article-title': 'Binaryconnect: Training deep neural networks with binary weights during propagations', 'author': 'courbariaux', 'year': '2015', 'journal-title': 'NIPS'} / NIPS / Binaryconnect: Training deep neural networks with binary weights during propagations by courbariaux (2015)
{'key': 'ref6', 'first-page': '1269', 'article-title': 'Exploiting linear structure within convolutional networks for efficient evaluation', 'author': 'denton', 'year': '2014', 'journal-title': 'NIPS'} / NIPS / Exploiting linear structure within convolutional networks for efficient evaluation by denton (2014)
{'key': 'ref29', 'first-page': '1310', 'article-title': 'On the difficulty of training recurrent neural networks', 'author': 'pascanu', 'year': '2013', 'journal-title': 'ICML'} / ICML / On the difficulty of training recurrent neural networks by pascanu (2013)
{'key': 'ref5', 'first-page': '2148', 'article-title': 'Predicting parameters in deep learning', 'author': 'denil', 'year': '2013', 'journal-title': 'NIPS'} / NIPS / Predicting parameters in deep learning by denil (2013)
{'key': 'ref8', 'volume': 'abs 1412 6115', 'author': 'gong', 'year': '2014', 'journal-title': 'Compressing deep convolutional networks using vector quantization'} / Compressing deep convolutional networks using vector quantization by gong (2014)
10.1109/ICCV.2015.169
{'key': 'ref2', 'first-page': '2285', 'article-title': 'Compressing neural networks with the hashing trick', 'author': 'chen', 'year': '2015', 'journal-title': 'ICML'} / ICML / Compressing neural networks with the hashing trick by chen (2015)
{'key': 'ref9', 'volume': 'abs 1510 149', 'author': 'han', 'year': '2015', 'journal-title': 'Deep compression Compressing deep neural network with pruning trained quantization and huffman coding'} / Deep compression Compressing deep neural network with pruning trained quantization and huffman coding by han (2015)
{'key': 'ref1', 'first-page': '354', 'article-title': 'A unified multi-scale deep convolutional neural network for fast object detection', 'author': 'cai', 'year': '2016', 'journal-title': 'ECCV'} / ECCV / A unified multi-scale deep convolutional neural network for fast object detection by cai (2016)
{'year': '2009', 'author': 'krizhevsky', 'journal-title': 'Learning multiple layers of features from tiny images', 'key': 'ref20'} / Learning multiple layers of features from tiny images by krizhevsky (2009)
{'key': 'ref22', 'volume': 'abs 1605 4711', 'author': 'li', 'year': '2016', 'journal-title': 'Ternary Weight Networks'} / Ternary Weight Networks by li (2016)
{'key': 'ref21', 'first-page': '1106', 'article-title': 'Imagenet classification with deep convolutional neural networks', 'author': 'krizhevsky', 'year': '2012', 'journal-title': 'NIPS'} / NIPS / Imagenet classification with deep convolutional neural networks by krizhevsky (2012)
{'key': 'ref24', 'first-page': '2849', 'article-title': 'Fixed point quantization of deep convolutional networks', 'author': 'lin', 'year': '2016', 'journal-title': 'ICML'} / ICML / Fixed point quantization of deep convolutional networks by lin (2016)
{'key': 'ref23', 'volume': 'abs 1607 2241', 'author': 'lin', 'year': '2016', 'journal-title': 'Overcoming challenges in fixed point training of deep convolutional networks'} / Overcoming challenges in fixed point training of deep convolutional networks by lin (2016)
{'key': 'ref26', 'volume': 'abs 1510 3009', 'author': 'lin', 'year': '2015', 'journal-title': 'Neural networks with few multiplications'} / Neural networks with few multiplications by lin (2015)
{'key': 'ref25', 'volume': 'abs 1312 4400', 'author': 'lin', 'year': '2013', 'journal-title': 'Network in Network'} / Network in Network by lin (2013)

Dates

Type	When
Created	7 years, 9 months ago (Nov. 9, 2017, 4:50 p.m.)
Deposited	7 years, 8 months ago (Dec. 18, 2017, 5:22 p.m.)
Indexed	1 week, 1 day ago (Aug. 30, 2025, 1:16 p.m.)
Issued	8 years, 2 months ago (July 1, 2017)
Published	8 years, 2 months ago (July 1, 2017)
Published Print	8 years, 2 months ago (July 1, 2017)

Funders 0

None

BibTeX

@inproceedings{Cai_2017, title={Deep Learning with Low Precision by Half-Wave Gaussian Quantization}, url={http://dx.doi.org/10.1109/cvpr.2017.574}, DOI={10.1109/cvpr.2017.574}, booktitle={2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, publisher={IEEE}, author={Cai, Zhaowei and He, Xiaodong and Sun, Jian and Vasconcelos, Nuno}, year={2017}, month=jul, pages={5406–5414} }

JSON

{
  "indexed": {
    "date-parts": [
      [
        2025,
        8,
        30
      ]
    ],
    "date-time": "2025-08-30T17:16:35Z",
    "timestamp": 1756574195812,
    "version": "3.28.0"
  },
  "reference-count": 37,
  "publisher": "IEEE",
  "content-domain": {
    "domain": [],
    "crossmark-restriction": false
  },
  "published-print": {
    "date-parts": [
      [
        2017,
        7
      ]
    ]
  },
  "DOI": "10.1109/cvpr.2017.574",
  "type": "proceedings-article",
  "created": {
    "date-parts": [
      [
        2017,
        11,
        9
      ]
    ],
    "date-time": "2017-11-09T21:50:33Z",
    "timestamp": 1510264233000
  },
  "page": "5406-5414",
  "source": "Crossref",
  "is-referenced-by-count": 314,
  "title": "Deep Learning with Low Precision by Half-Wave Gaussian Quantization",
  "prefix": "10.1109",
  "author": [
    {
      "given": "Zhaowei",
      "family": "Cai",
      "sequence": "first",
      "affiliation": []
    },
    {
      "given": "Xiaodong",
      "family": "He",
      "sequence": "additional",
      "affiliation": []
    },
    {
      "given": "Jian",
      "family": "Sun",
      "sequence": "additional",
      "affiliation": []
    },
    {
      "given": "Nuno",
      "family": "Vasconcelos",
      "sequence": "additional",
      "affiliation": []
    }
  ],
  "member": "263",
  "reference": [
    {
      "key": "ref33",
      "volume": "abs 1409 1556",
      "author": "simonyan",
      "year": "2014",
      "journal-title": "Very Deep Convolutional Networks for Large-scale Image Recognition"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref32",
      "DOI": "10.1007/s11263-015-0816-y"
    },
    {
      "key": "ref31",
      "first-page": "91",
      "article-title": "Faster R-CNN: towards real-time object detection with region proposal networks",
      "author": "ren",
      "year": "2015",
      "journal-title": "NIPS"
    },
    {
      "key": "ref30",
      "first-page": "525",
      "article-title": "Xnornet: Imagenet classification using binary convolutional neural networks",
      "author": "rastegari",
      "year": "2016",
      "journal-title": "ECCV"
    },
    {
      "key": "ref37",
      "volume": "abs 1606 6160",
      "author": "zhou",
      "year": "2016",
      "journal-title": "DoReFa-Net Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref36",
      "DOI": "10.1109/ICASSP.2013.6638312"
    },
    {
      "key": "ref35",
      "article-title": "Improving the speed of neural networks on cpus",
      "author": "vanhoucke",
      "year": "2011",
      "journal-title": "Proc Deep Learning and Unsupervised Feature Learning Workshop NIPS 2011"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref34",
      "DOI": "10.1109/CVPR.2015.7298594"
    },
    {
      "key": "ref10",
      "first-page": "1135",
      "article-title": "Learning both weights and connections for efficient neural network",
      "author": "han",
      "year": "2015",
      "journal-title": "NIPS"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref11",
      "DOI": "10.1109/ICCV.2015.123"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref12",
      "DOI": "10.1109/CVPR.2016.90"
    },
    {
      "key": "ref13",
      "volume": "abs 1207 580",
      "author": "hinton",
      "year": "2012",
      "journal-title": "Improving Neural Networks by Preventing Co-adaptation of Feature Detectors"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref14",
      "DOI": "10.1214/aoms/1177703732"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref15",
      "DOI": "10.1016/S0893-6080(00)00026-5"
    },
    {
      "key": "ref16",
      "volume": "abs 1602 7360",
      "author": "iandola",
      "year": "2016",
      "journal-title": "SqueezeNet AlexNet-level accuracy with 50x fewer parameters and < 1MB model size"
    },
    {
      "key": "ref17",
      "first-page": "448",
      "article-title": "Batch normalization: Accelerating deep network training by reducing internal covariate shift",
      "author": "ioffe",
      "year": "2015",
      "journal-title": "ICML"
    },
    {
      "key": "ref18",
      "article-title": "Speeding up convolutional neural networks with low rank expansions",
      "author": "jaderberg",
      "year": "2014",
      "journal-title": "BMVC"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref19",
      "DOI": "10.1145/2647868.2654889"
    },
    {
      "key": "ref28",
      "first-page": "807",
      "article-title": "Rectified linear units improve restricted boltzmann machines",
      "author": "nair",
      "year": "2010",
      "journal-title": "ICML"
    },
    {
      "key": "ref4",
      "volume": "abs 1602 2830",
      "author": "courbariaux",
      "year": "2016",
      "journal-title": "Binarized neural networks Training neural networks with weights and activations constrained to +1 or ?1"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref27",
      "DOI": "10.1109/TIT.1982.1056489"
    },
    {
      "key": "ref3",
      "first-page": "3123",
      "article-title": "Binaryconnect: Training deep neural networks with binary weights during propagations",
      "author": "courbariaux",
      "year": "2015",
      "journal-title": "NIPS"
    },
    {
      "key": "ref6",
      "first-page": "1269",
      "article-title": "Exploiting linear structure within convolutional networks for efficient evaluation",
      "author": "denton",
      "year": "2014",
      "journal-title": "NIPS"
    },
    {
      "key": "ref29",
      "first-page": "1310",
      "article-title": "On the difficulty of training recurrent neural networks",
      "author": "pascanu",
      "year": "2013",
      "journal-title": "ICML"
    },
    {
      "key": "ref5",
      "first-page": "2148",
      "article-title": "Predicting parameters in deep learning",
      "author": "denil",
      "year": "2013",
      "journal-title": "NIPS"
    },
    {
      "key": "ref8",
      "volume": "abs 1412 6115",
      "author": "gong",
      "year": "2014",
      "journal-title": "Compressing deep convolutional networks using vector quantization"
    },
    {
      "doi-asserted-by": "publisher",
      "key": "ref7",
      "DOI": "10.1109/ICCV.2015.169"
    },
    {
      "key": "ref2",
      "first-page": "2285",
      "article-title": "Compressing neural networks with the hashing trick",
      "author": "chen",
      "year": "2015",
      "journal-title": "ICML"
    },
    {
      "key": "ref9",
      "volume": "abs 1510 149",
      "author": "han",
      "year": "2015",
      "journal-title": "Deep compression Compressing deep neural network with pruning trained quantization and huffman coding"
    },
    {
      "key": "ref1",
      "first-page": "354",
      "article-title": "A unified multi-scale deep convolutional neural network for fast object detection",
      "author": "cai",
      "year": "2016",
      "journal-title": "ECCV"
    },
    {
      "year": "2009",
      "author": "krizhevsky",
      "journal-title": "Learning multiple layers of features from tiny images",
      "key": "ref20"
    },
    {
      "key": "ref22",
      "volume": "abs 1605 4711",
      "author": "li",
      "year": "2016",
      "journal-title": "Ternary Weight Networks"
    },
    {
      "key": "ref21",
      "first-page": "1106",
      "article-title": "Imagenet classification with deep convolutional neural networks",
      "author": "krizhevsky",
      "year": "2012",
      "journal-title": "NIPS"
    },
    {
      "key": "ref24",
      "first-page": "2849",
      "article-title": "Fixed point quantization of deep convolutional networks",
      "author": "lin",
      "year": "2016",
      "journal-title": "ICML"
    },
    {
      "key": "ref23",
      "volume": "abs 1607 2241",
      "author": "lin",
      "year": "2016",
      "journal-title": "Overcoming challenges in fixed point training of deep convolutional networks"
    },
    {
      "key": "ref26",
      "volume": "abs 1510 3009",
      "author": "lin",
      "year": "2015",
      "journal-title": "Neural networks with few multiplications"
    },
    {
      "key": "ref25",
      "volume": "abs 1312 4400",
      "author": "lin",
      "year": "2013",
      "journal-title": "Network in Network"
    }
  ],
  "event": "2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)",
  "container-title": "2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)",
  "original-title": [],
  "link": [
    {
      "URL": "http://xplorestaging.ieee.org/ielx7/8097368/8099483/08100057.pdf?arnumber=8100057",
      "content-type": "unspecified",
      "content-version": "vor",
      "intended-application": "similarity-checking"
    }
  ],
  "deposited": {
    "date-parts": [
      [
        2017,
        12,
        18
      ]
    ],
    "date-time": "2017-12-18T22:22:33Z",
    "timestamp": 1513635753000
  },
  "score": 1,
  "resource": {
    "primary": {
      "URL": "http://ieeexplore.ieee.org/document/8100057/"
    }
  },
  "subtitle": [],
  "short-title": [],
  "issued": {
    "date-parts": [
      [
        2017,
        7
      ]
    ]
  },
  "references-count": 37,
  "URL": "http://dx.doi.org/10.1109/CVPR.2017.574",
  "relation": {},
  "subject": [],
  "published": {
    "date-parts": [
      [
        2017,
        7
      ]
    ]
  }
}