ETRI Journal

Electronics and Telecommunications Research Institute (한국전자통신연구원)
권호 표지
DOI QR코드

DOI QR Code

Interworking technology of neural network and data among deep learning frameworks

  • Park, Jaebok (Artificial Intelligence Research Laboratory, Electronics and Telecommunications Research Institute) ;
  • Yoo, Seungmok (Artificial Intelligence Research Laboratory, Electronics and Telecommunications Research Institute) ;
  • Yoon, Seokjin (Artificial Intelligence Research Laboratory, Electronics and Telecommunications Research Institute) ;
  • Lee, Kyunghee (Artificial Intelligence Research Laboratory, Electronics and Telecommunications Research Institute) ;
  • Cho, Changsik (Artificial Intelligence Research Laboratory, Electronics and Telecommunications Research Institute)
  • Received : 2018.04.23
  • Accepted : 2019.05.20
  • Published : 2019.12.06
Download PDF
⟨ Previous Next ⟩

Abstract

Based on the growing demand for neural network technologies, various neural network inference engines are being developed. However, each inference engine has its own neural network storage format. There is a growing demand for standardization to solve this problem. This study presents interworking techniques for ensuring the compatibility of neural networks and data among the various deep learning frameworks. The proposed technique standardizes the graphic expression grammar and learning data storage format using the Neural Network Exchange Format (NNEF) of Khronos. The proposed converter includes a lexical, syntax, and parser. This NNEF parser converts neural network information into a parsing tree and quantizes data. To validate the proposed system, we verified that MNIST is immediately executed by importing AlexNet's neural network and learned data. Therefore, this study contributes an efficient design technique for a converter that can execute a neural network and learned data in various frameworks regardless of the storage format of each framework.

Keywords

References

  1. Khronos, Neural network exchange format (NNEF) specification, Beaverton, OR, USA, 2018.
  2. H. Deepika, N. Mangala, and S.C. Babu, Automatic program generation for heterogeneous architectures, in Proc. Int. Adv. Comput., Commun. Inf. Conf., Jaipur, India, Sept. 2016, pp. 102-109.
  3. A. Krizhevsky, I. Sutskever, and G.E. Hinton, Imagenet classification with deep convolutional neural networks, in Proc. Int. Neural Inf. Process. Syst. Conf., Lake Tahoe, NV, USA, Dec. 2012, pp. 1097-1105.
  4. K. Simonyan and A. Zisserman, Very deep convolutional networks for large-scale image recognition, in Proc. Int. Learn. Representations Conf., San Diego, CA, USA, May 2015, pp. 1-14.
  5. Y. LeCun, Y. Bengio, and G. Hinton, Deep learning, Nature J. 521 (2015), 436-444. https://doi.org/10.1038/nature14539
  6. J. Deng et al., Imagenet: A large-scale hierarchical image database, in Proc. Int. IEEE Comput. Vision Pattern Recogn. Conf., Miami, FL, USA, June 2009, pp. 248-255.
  7. M. Abadi et al., TensorFlow: Large-scale machine learning on heterogeneous distributed systems, in Proc. Int. USENIX Symp. Oper. Syst. Des. Implement., Savannah, GA, USA, 2016, pp. 265-283.
  8. F. Chollet, Deep learning with python, Manning Publications, Shelter Island, NY, 2017.
  9. Y. Jia et al., Caffe: Convolutional architecture for fast feature embedding, in Proc. Int. ACM Multimedia Conf., Orlando, FL, USA, Nov. 2014, pp. 675-678.
  10. I. Serrano et al., Fight recognition in video using hough forests and 2D convolutional neural network, IEEE Trans. Image Process. 27 (2018), no. 10, 4787-4797. https://doi.org/10.1109/TIP.2018.2845742
  11. Facebook Open Source, Caffe2: a new lightweight, modular, and scalable deep learning framework, Accessed Jan. 2018, http://caffe2.ai
  12. R. Collobert, K. Kavukcuoglu, and C. Farabet, Torch7: A MATLAB-like environment for machine learning, BigLearn, NIPS Workshop, no. EPFL-CONF-192376, 2011.
  13. H. Adie, I. Paradana, and Pranowo, Parallel computing accelerated image inpainting using GPU CUDA, Theano, and Tensorflow, in Proc Int. Inf. Technol. Electr. Eng., Kuta, Indonesia, July 2018, pp. 621-625.
  14. S. Chetlur et al., CUDNN: Efficient primitives for deep learning, CoRR (2014). http://arxiv.org/abs/1410.0759
  15. T.M. Breuel, High performance text recognition using a hybrid convolutional-LSTM implementation, in Proc. Int. IEEE IAPR Document Anal. Recogn., Kyoto, Japan, Nov. 2017, pp. 11-16.
  16. The Theano Development Team, Theano: a python framework for fast computation of mathematical expressions, arXiv preprint, 2016, arXiv:1605.02688.
  17. F. Chollet, Keras: deep learning library for Theano and Tensorflow, GitHub Repository, Tech. Rep., 2015, https://keras.io/getting-started/faq/#how-should-i-cite-keras
  18. H. Kim et al., Applied machine learning at Facebook: a datacenter infrastructure perspective, in Proc. Int. IEEE Symp. High Performance Comput. Architecture, Vienna, Austria, Feb. 2018, pp. 620-629.
  19. J.Q. Candela, Facebook and Microsoft introduce new open ecosystem for interchangeable AI frameworks, Sept. 2017.
  20. J. Ambrosi et al., Hardware-software co-design for an analogdigital accelerator for machine learning, in Proc. Int. IEEE Rebooting Comput., Mclean, VA, USA, Nov. 2018, pp. 1-13.
  21. B. Seo et al., Top-down parsing for Neural Network Exchange Format (NNEF) in TensorFlow-based deep learning computation, in Proc. Int. Inf. Netw., Chiang Mai, Thailand, Jan. 2018, pp. 522-524.
  22. M. Shin et al., Neural network syntax analyzer for embedded standardized deep learning, in Proc. Int. Workshop Embedded Mobile Deep Learn., Munich, Germany, June 2018, pp. 37-41.
  23. T. Chen et al., MXNet: A flexible and efficient machine learning library for heterogeneous distributed systems, arXiv preprint, 2015, arXiv:1512.01274.
  24. J. Hanhirova et al., Latency and throughput characterization of convolutional neural networks for mobile computer vision, in Proc. Int. ACM Multimedia Syst. Conf., Amsterdam, Netherlands, June 2018, pp. 204-215.
  25. Z. Guo, J. Han, and T. Li, Implementing OpenVX on a polymorphous array processor, in Proc. Int. IEEE Commun. Technol. Conf., Hangzhou, China, Oct. 2015, pp. 598-601.
  26. M. Upadhyaya, Simple calculator compiler using Lex and YACC, in Proc. Int. IEEE Electron. Comput. Technol. Conf., Kanyakumari, India, Apr. 2011, pp. 182-187.
  27. P. Nakwijit and P. Ratanaworabhan, A parser generator using the Grammar Flow Graph, in Proc. Int. IEEE Comput. Sci. Eng. Conf., Chiang Mai, Thiland, Nov. 2015, pp. 1-6.

Cited by

  1. Spiking Neural Networks(SNN) 구조에서 뉴런의 개수와 학습량에 따른 학습 성능 변화 분석 vol.6, pp.3, 2019, https://doi.org/10.17703/jcct.2020.6.3.463