Page 84 - ITU Journal, ICT Discoveries, Volume 3, No. 1, June 2020 Special issue: The future of video and immersive media

ITU Journal: ICT Discoveries, Vol. 3(1), June 2020

[61] D. B. Skillicorn. Foundations of parallel program- [74] D. Wang, A. Khosla, R. Gargeya, H. Irshad,
ming. Number 6. Cambridge University Press, and A. H. Beck. Deep learning for identify-
2005. ing metastatic breast cancer. arXiv preprint
arXiv:1606.05718, 2016.
[62] S. U. Stich. Local sgd converges fast and communi-
cates little. arXiv preprint arXiv:1805.09767, 2018. [75] H. Wang, S. Sievert, S. Liu, Z. Charles, D. Papail-
iopoulos, and S. Wright. Atomo: Communication-
[63] S. U. Stich, J.-B. Cordonnier, and M. Jaggi. Sparsi- efficient learning via atomic sparsification. In Ad-
fied sgd with memory. In Advances in Neural Infor- vances in Neural Information Processing Systems,
mation Processing Systems, pages 4447–4458, 2018. pages 9850–9861, 2018.

[64] P. Subramanyan, R. Sinha, I. Lebedev, S. Devadas, [76] J. Wang and G. Joshi. Cooperative sgd: A
and S. A. Seshia. A formal foundation for secure unified framework for the design and analysis
remote execution of enclaves. In Proceedings of the of communication-efficient sgd algorithms. arXiv
2017 ACM SIGSAC Conference on Computer and preprint arXiv:1808.07576, 2018.
Communications Security, pages 2435–2450. ACM,
2017. [77] N. Wang, J. Choi, D. Brand, C.-Y. Chen, and
K. Gopalakrishnan. Training deep neural networks
[65] I. Sutskever, O. Vinyals, and Q. V. Le. Sequence with 8-bit floating point numbers. In Advances in
to sequence learning with neural networks. In Ad- neural information processing systems, pages 7675–
vances in neural information processing systems, 7684, 2018.
pages 3104–3112, 2014.
[78] W. Wen, C. Xu, F. Yan, C. Wu, Y. Wang, Y. Chen,
[66] H. Tang, X. Lian, S. Qiu, L. Yuan, C. Zhang, and H. Li. Terngrad: Ternary gradients to reduce
T. Zhang, and J. Liu. Deepsqueeze: Paral- communication in distributed deep learning. arXiv
lel stochastic gradient descent with double-pass preprint arXiv:1705.07878, 2017.
error-compensated compression. arXiv preprint [79] S. Wiedemann, H. Kirchoffer, S. Matlage, P. Haase,
arXiv:1907.07346, 2019. A. Marban, T. Marinc, D. Neumann, T. Nguyen,
A. Osman, D. Marpe, et al. Deepcabac: A univer-
[67] H. Tang, X. Lian, M. Yan, C. Zhang, and J. Liu. D sal compression algorithm for deep neural networks.
^2: Decentralized training over decentralized data. arXiv preprint arXiv:1907.11900, 2019.
arXiv preprint arXiv:1803.07068, 2018.
[80] S. Wiedemann, A. Marban, K.-R. Müller, and
[68] A. Tjandra, S. Sakti, and S. Nakamura. Tensor de- W. Samek. Entropy-constrained training of deep
composition for compressing recurrent neural net- neural networks. In 2019 International Joint Con-
work. In 2018 International Joint Conference on ference on Neural Networks (IJCNN), pages 1–8.
Neural Networks (IJCNN), pages 1–8. IEEE, 2018.
IEEE, 2019.
[69] M. Tu, V. Berisha, Y. Cao, and J.-s. Seo. Reduc- [81] S. Wiedemann, K.-R. Müller, and W. Samek. Com-
ing the model order of deep neural networks using pact and computationally efficient representation of
information theory. In 2016 IEEE Computer So- deep neural networks. IEEE Transactions on Neu-
ciety Annual Symposium on VLSI (ISVLSI), pages ral Networks and Learning Systems, 31(3):772–785,
93–98. IEEE, 2016. 2020.

[70] P. Vanhaesebrouck, A. Bellet, and M. Tommasi. [82] B. Wu, X. Dai, P. Zhang, Y. Wang, F. Sun, Y. Wu,
Decentralized collaborative learning of personalized Y. Tian, P. Vajda, Y. Jia, and K. Keutzer. Fbnet:
models over networks. 2017. Hardware-aware efficient convnet design via differ-
entiable neural architecture search. In Proceedings
[71] D. R. Varma. Managing dicom images: Tips and of the IEEE Conference on Computer Vision and
tricks for the radiologist. The Indian journal of Pattern Recognition, pages 10734–10742, 2019.
radiology & imaging, 22(1):4, 2012.
[83] K. Xu, J. Ba, R. Kiros, K. Cho, A. Courville,
[72] T. Vogels, S. P. Karimireddy, and M. Jaggi. R. Salakhudinov, R. Zemel, and Y. Bengio. Show,
Powersgd: Practical low-rank gradient compres- attend and tell: Neural image caption generation
sion for distributed optimization. arXiv preprint with visual attention. In International Conference
arXiv:1905.13727, 2019. on Machine Learning, pages 2048–2057, 2015.
[73] P. Voigt and A. Von dem Bussche. The eu gen- [84] T.-J. Yang, Y.-H. Chen, and V. Sze. Designing
eral data protection regulation (gdpr). A Practical energy-efficient convolutional neural networks us-
Guide, 1st Ed., 2017. ing energy-aware pruning. In Proceedings of the

62 © International Telecommunication Union, 2020