近年來，善用邊緣運算能力的攝影機 (以下稱為邊緣攝影機) 結合深度神經網路來實現人工智慧物聯網 (AIoT) 的應用不斷增加，讓具備「低接觸服務」的「無人商店」之智慧生活得以實現。然而，在無人商店佈署大量邊緣攝影機並訓練其邊緣模型 (edge model) 非常消耗時間與人力成本。因此，已有研究提出實例轉移學習 (transfer learning) 方法，但其訓練邊緣模型都需要強大伺服器的協助。雖有研究提出不需要伺服器的直接端對端實例轉移，但其除了無法發揮潛藏特徵之效益，傳輸上仍需要較大的頻寬，且占用較長的模型訓練時間，甚至造成隱私外洩的風險。此外，既有研究需人力介入來協助新邊緣攝影機的加入，也不符合物聯網降低人為介入的特性。因此，本研究提出「直接端對端一對多潛藏特徵轉移學習技術」，其讓主要一台邊緣攝影機作為多台邊緣攝影機的潛藏特徵轉移來源，除了提高知識的重用度，並加速初始模型的建立。另外，為了發揮邊緣攝影機來源資料的多元性，本研究提出「直接端對端多對一潛藏特徵轉移學習技術」。此外，為了降低加入新邊緣攝影機的人為介入，本研究進一步提出支援服務發現機制的「直接端對端自主資訊交換轉移學習平台」，其透過去中心化傳輸協定來直接端對端傳送特徵資訊以降低隱私外洩的風險。基於「直接端對端自主資訊交換轉移學習平台」，實驗結果顯示，「直接端對端一對多潛藏特徵轉移學習技術」比既有研究平均提升6.30%的準確度，並比使用實例樣本之轉移學習在邊緣上直接訓練邊緣模型時間平均節省32.15%，且可在邊緣上平均節省傳送22.92%的傳輸成本，最高更可節省高達 83.33%的傳輸成本。另外，「直接端對端多對一潛藏特徵轉移學習技術」也比既有研究平均提升3.42%的準確度，並比使用實例樣本之轉移學習在邊緣上直接訓練邊緣模型時間平均節省66.99%，且可在邊緣上平均節省56.67%的傳輸成本，充分證明本研究提出方法的有效性與可應用性。

There has been an increasing number of smart cameras that leverage edge-computing (hereafter referred to as edge cameras) combined with deep neural networks to realize the Artificial Intelligence of Things (AIoT), enabling a smart life with "low-touch services," such as unmanned stores. However, deploying a large number of edge cameras and training their models (aka edge models) in unmanned stores are very time-consuming and labor-intensive. Therefore, studies have utilized transfer learning methods, but training edge models often requires the help of powerful servers. Although a study has proposed direct edge-to-edge instance transfer, it does not exploit latent features, still requiring larger bandwidth, taking longer training time, and causing privacy leakage. Therefore, we propose "Direct Edge-to-Edge One-to-Many Latent Feature Transfer Learning (De2eOMLFTL)", to allows one edge camera to be the latent-feature source of multiple edge cameras, which not only improves knowledge reuse but also accelerates initial model training. In addition, in order to exploit the diversity of source edge cameras, we propose "Direct Edge-to-Edge Many-to-One Latent Feature Transfer Learning (De2eMOLFTL)." Since the existing research requires human intervention to assist the joining of new edge cameras. we further propose an "Direct Edge-to-Edge Autonomous Information Exchange Platform (De2eAIEP)" to reduce the human intervention. Based on the De2eAIEP, the experimental results show that the De2eOMLFTL improves the accuracy by 6.30%, saves 32.15% on training time, and saves 22.92% (maximum 83.33%) on transmission cost. In addition, the accuracy of the De2eMOLFTL increases by 3.42%, and saves 66.99% on training time, and 56.67% on transmission cost.

[1] A. Banafa. (2019, November 21). Ten Trends of IoT in 2020. Available: https://iot.ieee.org/newsletter/november-2019/ten-trends-of-iot-in-2020
[2] S. Pouyanfar et al., "A survey on deep learning: Algorithms, techniques, and applications," ACM computing surveys (CSUR), vol. 51, no. 5, p. 92, 2019.
[3] J. Hu, L. Shen, S. Albanie, G. Sun, and E. Wu, "Squeeze-and-Excitation Networks," ArXiv e-prints, Accessed on: September 01, 2017Available: https://ui.adsabs.harvard.edu/#abs/2017arXiv170901507H
[4] G. Ananthanarayanan et al., "Real-time video analytics: The killer app for edge computing," computer, vol. 50, no. 10, pp. 58-67, 2017.
[5] N. Garun, "Amazon just launched a cashier-free convenience store," The Verge.(Retrieved April 17, 2017 from http://www. theverge. com/2016/12/5/13842592/amazongo-new-cashier-less-convenience-store), 2016.
[6] F. Zhuang et al., "A comprehensive survey on transfer learning," Proceedings of the IEEE, 2020.
[7] Z. Tao and Q. Li, "esgd: Communication efficient distributed deep learning on the edge," in {USENIX} Workshop on Hot Topics in Edge Computing (HotEdge 18), 2018.
[8] R. Sharma, S. Biookaghazadeh, B. Li, and M. Zhao, "Are existing knowledge transfer techniques effective for deep learning with edge devices?," in 2018 IEEE International Conference on Edge Computing (EDGE), 2018, pp. 42-49: IEEE.
[9] C.-H. Lu and X.-Z. Lin, "Towards Direct Edge-to-edge Transfer Learning for IoT-enabled Edge Cameras," IEEE IoT Journal, 2020 (Online Published).
[10] F. A. Alaba, M. Othman, I. A. T. Hashem, and F. Alotaibi, "Internet of Things security: A survey," Journal of Network and Computer Applications, vol. 88, pp. 10-28, 2017.
[11] S. Ruder, "An Overview of Multi-Task Learning in Deep Neural Networks," ArXiv e-prints, vol. 1706, Accessed on: June 1, 2017Available: http://adsabs.harvard.edu/abs/2017arXiv170605098R
[12] A. Bettge, R. Roscher, and S. Wenzel, "Deep Self-taught Learning for Remote Sensing Image Classification," CoRR, vol. abs/1710.07096, 2017.
[13] Y. Ganin et al., "Domain-Adversarial Training of Neural Networks," Journal of Machine Learning Research, vol. 17, pp. 59:1-59:35, 2016.
[14] M. Johnson et al., "Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation," TACL, vol. 5, pp. 339-351, 2017.
[15] W. Dai, Q. Yang, G.-R. Xue, and Y. Yu, "Self-taught clustering," in ICML, 2008.
[16] P. Sinno Jialin and Y. Qiang, "A Survey on Transfer Learning," Knowledge and Data Engineering, IEEE Transactions on, vol. 22, no. 10, pp. 1345-1359, 2010.
[17] E. Guttman, C. Perkins, J. Veizades, and M. Day, "Service location protocol, version 2," ed: RFC 2608, June, 1999.
[18] A. Presser, L. Farrell, D. Kemp, and W. Lupton, "Upnp device architecture 1.1," in UPnP Forum, 2008, vol. 22.
[19] M. A. El Saoud, T. Kunz, and S. Mahmoud, "SLPManet: service location protocol for MANET," in Proceedings of the 2006 international conference on Wireless communications and mobile computing, 2006, pp. 701-706.
[20] E. Guttman, "Service location protocol: Automatic discovery of IP network services," IEEE Internet Computing, vol. 3, no. 4, pp. 71-80, 1999.
[21] B. C. Villaverde, R. D. P. Alberola, A. J. Jara, S. Fedor, S. K. Das, and D. Pesch, "Service discovery protocols for constrained machine-to-machine communications," IEEE communications surveys & tutorials, vol. 16, no. 1, pp. 41-60, 2013.
[22] S. Cheshire and M. Krochmal, "Multicast dns," RFC 6762, February2013.
[23] S. Cheshire and M. Krochmal, "DNS-based service discovery," RFC 6763, February2013.
[24] C. Severance, "Roy t. fielding: Understanding the rest style," Computer, vol. 48, no. 6, pp. 7-9, 2015.
[25] W. Shang, Y. Yu, R. Droms, and L. Zhang, "Challenges in IoT networking via TCP/IP architecture," Technical Report NDN-0038. NDN Project, 2016.
[26] J. Dizdarević, F. Carpio, A. Jukan, and X. Masip-Bruin, "A survey of communication protocols for internet of things and related challenges of fog and cloud computing integration," ACM Computing Surveys (CSUR), vol. 51, no. 6, pp. 1-29, 2019.
[27] Y. Xu, V. Mahendran, W. Guo, and S. Radhakrishnan, "Fairness in fog networks: Achieving fair throughput performance in MQTT-based IoTs," in 2017 14th IEEE Annual Consumer Communications & Networking Conference (CCNC), 2017, pp. 191-196: IEEE.
[28] M. Sugiyama, S. Nakajima, H. Kashima, P. V. Buenau, and M. Kawanabe, "Direct importance estimation with model selection and its application to covariate shift adaptation," in Advances in neural information processing systems, 2008, pp. 1433-1440.
[29] J. Davis and P. Domingos, "Deep transfer via second-order markov logic," in Proceedings of the 26th annual international conference on machine learning, 2009, pp. 217-224.
[30] J. Blitzer, R. McDonald, and F. Pereira, "Domain adaptation with structural correspondence learning," in Proceedings of the 2006 conference on empirical methods in natural language processing, 2006, pp. 120-128.
[31] J. Gao, W. Fan, J. Jiang, and J. Han, "Knowledge transfer via multiple model local structure mapping," in Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, 2008, pp. 283-291.
[32] Z. Chen, J. Zhuang, X. Liang, and L. Lin, "Blending-Target Domain Adaptation by Adversarial Meta-Adaptation Networks," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 2248-2257.
[33] W. Ying, Y. Zhang, J. Huang, and Q. Yang, "Transfer learning via learning to transfer," in International Conference on Machine Learning, 2018, pp. 5072-5081.
[34] Y. Zhu, F. Zhuang, and D. Wang, "Aligning domain-specific distribution and classifier for cross-domain classification from multiple sources," in Proceedings of the AAAI Conference on Artificial Intelligence, 2019, vol. 33, pp. 5989-5996.
[35] Q. Wu, X. Zhou, Y. Yan, H. Wu, and H. Min, "Online transfer learning by leveraging multiple source domains," Knowledge and Information Systems, journal article vol. 52, no. 3, pp. 687-707, September 01 2017.
[36] Y. Freund and R. E. Schapire, "A decision-theoretic generalization of on-line learning and an application to boosting," Journal of computer and system sciences, vol. 55, no. 1, pp. 119-139, 1997.
[37] D. Zhang, X. Chen, D. Wang, and J. Shi, "A Survey on Collaborative Deep Learning and Privacy-Preserving," in 2018 IEEE Third International Conference on Data Science in Cyberspace (DSC), 2018, pp. 652-658: IEEE.
[38] K. Lin, S. Wang, and J. Zhou, "Collaborative deep reinforcement learning," arXiv preprint arXiv:1702.05796, 2017.
[39] D. Benditkis, A. Keren, L. Mor-Yosef, T. Avidor, N. Shoham, and N. Tal-Israel, "Distributed deep neural network training on edge devices," in Proceedings of the 4th ACM/IEEE Symposium on Edge Computing, 2019, pp. 304-306: ACM.
[40] J. Konečný, H. B. McMahan, F. X. Yu, P. Richtárik, A. T. Suresh, and D. Bacon, "Federated learning: Strategies for improving communication efficiency," arXiv preprint arXiv:1610.05492, 2016.
[41] T. Xing, S. S. Sandha, B. Balaji, S. Chakraborty, and M. Srivastava, "Enabling Edge Devices that Learn from Each Other: Cross Modal Training for Activity Recognition," in Proceedings of the 1st International Workshop on Edge Systems, Analytics and Networking, 2018, pp. 37-42: ACM.
[42] S. A. Rokni and H. Ghasemzadeh, "Autonomous Training of Activity Recognition Algorithms in Mobile Sensors: A Transfer Learning Approach in Context-Invariant Views," IEEE Transactions on Mobile Computing, 2018.
[43] O. Valery, P. Liu, and J. Wu, "CPU/GPU Collaboration Techniques for Transfer Learning on Mobile Devices," in 2017 IEEE 23rd International Conference on Parallel and Distributed Systems (ICPADS), 2017, pp. 477-484.
[44] H. Daga, P. K. Nicholson, A. Gavrilovska, and D. Lugones, "Cartel: A System for Collaborative Transfer Learning at the Edge," in Proceedings of the ACM Symposium on Cloud Computing, 2019, pp. 25-37.
[45] M. Ghifary, W. B. Kleijn, and M. Zhang, "Domain adaptive neural networks for object recognition," in Pacific Rim international conference on artificial intelligence, 2014, pp. 898-904: Springer.
[46] E. Tzeng, J. Hoffman, N. Zhang, K. Saenko, and T. Darrell, "Deep domain confusion: Maximizing for domain invariance," arXiv preprint arXiv:1412.3474, 2014.
[47] M. Long, Y. Cao, J. Wang, and M. I. Jordan, "Learning transferable features with deep adaptation networks," arXiv preprint arXiv:1502.02791, 2015.
[48] J. Li, W. Wu, D. Xue, and P. Gao, "Multi-Source Deep Transfer Neural Network Algorithm," Sensors, vol. 19, no. 18, p. 3992, 2019.
[49] A. Corsaro, "The data distribution service tutorial," Technical Report 4.0, 2014.
[50] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, and M. Ayyash, "Internet of things: A survey on enabling technologies, protocols, and applications," IEEE communications surveys & tutorials, vol. 17, no. 4, pp. 2347-2376, 2015.
[51] M. J. Swain and D. H. Ballard, "Color indexing," International journal of computer vision, vol. 7, no. 1, pp. 11-32, 1991.
[52] P. Chamoso, A. Rivas, J. J. Martín-Limorti, and S. Rodríguez, "A hash based image matching algorithm for social networks," in International Conference on Practical Applications of Agents and Multi-Agent Systems, 2017, pp. 183-190: Springer.
[53] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770-778.
[54] K. He, X. Zhang, S. Ren, and J. Sun, "Identity mappings in deep residual networks," in European conference on computer vision, 2016, pp. 630-645: Springer.
[55] H. Shimodaira, "Improving predictive inference under covariate shift by weighting the log-likelihood function," Journal of statistical planning and inference, vol. 90, no. 2, pp. 227-244, 2000.
[56] J. Huang, A. Gretton, K. Borgwardt, B. Schölkopf, and A. J. Smola, "Correcting sample selection bias by unlabeled data," in Advances in neural information processing systems, 2007, pp. 601-608.
[57] K. Yao, "Applications of reproducing kernel Hilbert spaces–bandlimited signal models," Information and Control, vol. 11, no. 4, pp. 429-444, 1967.
[58] B. Sun and K. Saenko, "Deep coral: Correlation alignment for deep domain adaptation," in European conference on computer vision, 2016, pp. 443-450: Springer.
[59] Y. Mansour, M. Mohri, and A. Rostamizadeh, "Domain adaptation with multiple sources," Advances in neural information processing systems, vol. 21, pp. 1041-1048, 2008.
[60] I. Hussain, Q. He, and Z. Chen, "AUTOMATIC FRUIT RECOGNITION BASED ON DCNN FOR COMMERCIAL SOURCE TRACE SYSTEM."
[61] J. Gama, I. Žliobaitė, A. Bifet, M. Pechenizkiy, and A. Bouchachia, "A survey on concept drift adaptation," ACM Computing Surveys (CSUR), vol. 46, no. 4, p. 44, 2014.
[62] Z. Zhang and M. Sabuncu, "Generalized cross entropy loss for training deep neural networks with noisy labels," in Advances in neural information processing systems, 2018, pp. 8778-8788.
[63] D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," arXiv preprint arXiv:1412.6980, 2014.