近年來，人工智慧領域的學者致力於改進機器學習與深度學習來解決過去其他領域中難以解決的問題，透過語音辨識、影像辨識與自然語言處理等技術解決人類生活中的各種瑣事已不再是難題；更進一步地說，人工智慧在智慧家庭與自駕車等領域已是不可動搖的基石，甚至是近年爆發的武漢肺炎，都能依靠人工智慧來進行快速篩檢，深度學習中影像辨識的貢獻功不可沒。
人工智慧看似是仙丹靈藥，但往往在遇到新的問題要解決時，就必須再次花費大量時間來進行調整與訓練，雖然專家們在分散式學習與遷移學習中進行了許多研究，但這還是不足以支撐真實世界龐大的資訊量；在數年前，Google提出了聯合學習這一概念，將蒐集的資料以不侵犯個資的方式進行聯盟式的學習與分享，進而將多節點資料完美地利用。借鏡聯合學習，我們提出一種具有良好隱私性，並且能有效利用多節點資源讓模型進行智慧演化的方法；我們利用零樣本學習將影像特徵轉換成人類容易識別的語義空間，再將語義空間透過子空間分類器來進行識別與發現新的類別，並將類別與參與智慧演化的夥伴分享，透過智慧演化的夥伴們互相分享與更新知識，形成具自主演化的智聯網。

Through speech recognition, image recognition, natural language processing and other technologies, it is no longer difficult to solve various problems in human life. In addition, artificial intelligence has become an unshakable cornerstone in the field of smart homes and autonomous vehicles. Even the outbreak of covid-19 in recent years can rely on artificial intelligence for rapid screening, and the contribution of image recognition in deep learning is indispensable.
Artificial intelligence seems to be a panacea, but when it comes to solving a new problem, it often takes a lot of time to adjust and train. Although experts have conducted a lot of research in distributed learning and transfer learning, this is still not enough to deal with the massive amount of information in the real world. A few years ago, Google put forward the concept of federated learning, using the collected data for learning and sharing it with workers, making perfect use of multi-node data without infringing on their privacy. Learned from federated learning, we propose a method that has good privacy and can effectively use multi-node resources for the intelligent evolution of the model. Use zero-shot learning to convert image features into a semantic space called CE (Class Embedding) that is easy for humans to recognize, and then use a subspace classifier to classify and discover new classes in CE. When a new class is discovered, share it with partners that participate in intelligent evolution. By sharing and updating intelligence with each other, a self-evolving IoI (Internet of Intelligence) is formed.

[1] B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, "Communication-efficient learning of deep networks from decentralized data," in Artificial intelligence and statistics, 2017: PMLR, pp. 1273-1282.
[2] S.-T. Cheng, C.-W. Hsu, G.-J. Horng, and C.-H. Lin, "Adaptive cache pre-forwarding policy for distributed deep learning," Computers & Electrical Engineering, vol. 82, p. 106558, 2020.
[3] A. Farhadi, I. Endres, D. Hoiem, and D. Forsyth, "Describing objects by their attributes," in 2009 IEEE conference on computer vision and pattern recognition, 2009: IEEE, pp. 1778-1785.
[4] C. H. Lampert, H. Nickisch, and S. Harmeling, "Learning to detect unseen object classes by between-class attribute transfer," in 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009: IEEE, pp. 951-958.
[5] C. H. Lampert, H. Nickisch, and S. Harmeling, "Attribute-based classification for zero-shot visual object categorization," IEEE transactions on pattern analysis and machine intelligence, vol. 36, no. 3, pp. 453-465, 2013.
[6] J. Lu, J. Li, Z. Yan, and C. Zhang, "Zero-shot learning by generating pseudo feature representations," arXiv preprint arXiv:1703.06389, 2017.
[7] Y. Xian, C. H. Lampert, B. Schiele, and Z. Akata, "Zero-shot learning—a comprehensive evaluation of the good, the bad and the ugly," IEEE transactions on pattern analysis and machine intelligence, vol. 41, no. 9, pp. 2251-2265, 2018.
[8] C.-C. Pan and S.-T. Chen, "Evolution of Class Embedding for Unseen Class in Zero-Shot Learning," 2021.
[9] T.-W. Pang and S.-T. Cheng, "P-Learning: Virtual Intelligence Evolution with Zero Pattern of Class Embedding," 2020.
[10] J. Nixon, J. Liu, and D. Berthelot, "Semi-supervised class discovery," arXiv preprint arXiv:2002.03480, 2020.
[11] C. L. Chamas, D. Cordeiro, and M. M. Eler, "Comparing REST, SOAP, Socket and gRPC in computation offloading of mobile applications: An energy cost analysis," in 2017 IEEE 9th Latin-American Conference on Communications (LATINCOM), 2017: IEEE, pp. 1-6.
[12] R. Biswas, X. Lu, and D. K. Panda, "Designing a micro-benchmark suite to evaluate gRPC for TensorFlow: Early experiences," arXiv preprint arXiv:1804.01138, 2018.
[13] R. Biswas, X. Lu, and D. K. Panda, "Accelerating tensorflow with adaptive rdma-based grpc," in 2018 IEEE 25th International Conference on High Performance Computing (HiPC), 2018: IEEE, pp. 2-11.
[14] M. Araújo, M. Maia, P. Rego, and J. de Souza, "Performance analysis of computational offloading on embedded platforms using the gRPC framework," in 8th International Workshop on ADVANCEs in ICT Infrastructures and Services (ADVANCE 2020), 2020, pp. 1--8.
[15] 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.
[16] C. Feichtenhofer, H. Fan, J. Malik, and K. He, "Slowfast networks for video recognition," in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 6202-6211.
[17] D. N. Osherson, J. Stern, O. Wilkie, M. Stob, and E. E. Smith, "Default probability," Cognitive Science, vol. 15, no. 2, pp. 251-269, 1991.
[18] C. Kemp, J. B. Tenenbaum, T. L. Griffiths, T. Yamada, and N. Ueda, "Learning systems of concepts with an infinite relational model," in AAAI, 2006, vol. 3, p. 5.
[19] L. Van der Maaten and G. Hinton, "Visualizing data using t-SNE," Journal of machine learning research, vol. 9, no. 11, 2008.