人工智慧不斷發展，衍生出機器學習與深度學習來解決過去其他領域中的難題，如語音辨識、影像辨識與自然語言處理等技術；而在人類生活中，視覺是非常重要的，因此在工廠自動化、自駕車、數位管家等問題中深度學習中影像辨識的貢獻功不可沒。
人工智慧看似無所不能，解決許多難題，但往往解決一個問題時，會需要類似經驗法則，也就是說需要大量的數據資料作為訓練,但是大量的數據資料蒐集，甚至將其整理成有用資料都是困難且耗時的。為了解決大量數據蒐集困難的問題，也為了比較符合接近人類的學習方式，學者們也紛紛投入小數據的研究，在此時，也有人提出了遷移式學習這概念，將從以前的任務當中去學習知識或經驗，應用於新的任務當中。換句話說，遷移學習目的是從一個或多個原本任務中抽取知識、經驗，然後應用於一個目標領域中,從中也產生有人將NLP技術引用進影像辨識中，例如Word Embedding賦予圖片有語義的結合方式。
利用遷移式學習與Word Embedding，我們提出一種新的分類器，該分類器由ensemble learning 將多個分類器組合而成之強分類器，組成強分類器的多個分類器由不同的 encoder 生成，此外，對於不同的 encoder，我們以不同 Word to vector model 所產生之 vector 訓練得出。

With the continuous development of artificial intelligence, machine learning and deep learning have been derived to solve problems in other fields in the past, such as speech recognition, image recognition and natural language processing. In the human environment, vision is very important, so the contribution of image recognition in deep learning for problems such as factory automation, self-driving cars, and digital housekeeping is indispensable.
Artificial intelligence seems to be omnipotent and solve many problems, but when solving a problem, it often requires similar rules of thumb, which means that a large amount of data is required for training, but collecting large amounts of data, or even sorting it into useful information, is difficult and time-consuming. In order to solve the problem of difficult collection of massive data, and in order to be more in line with human learning methods, experts are also committed to the research of small data. At this time, some people proposed the concept of transfer learning, which is to learn knowledge or experience from previous tasks and apply them to new tasks. In other words, the purpose of transfer learning is to extract knowledge and experience from one or more original tasks, and then apply it to a target field, from which some people refer to NLP technology into image recognition, such as Word Embedding to give images semantical combination method.
By leveraging transfer learning and word embedding, we propose a novel classifier that combines multiple classifiers via ensemble learning. different classifiers, which combined into a strong classifier, are generated by different encoders. In addition, we use vectors generated by different Word to Vector models to train different encoders.

Reference
[1] S. J. Pan and Q. Yang, "A Survey on Transfer Learning," IEEE Transactions on Knowledge and Data Engineering, 2010.
[2] Y. WANG, Q. Yao, J. T. Kwok, and L. M. Ni, "Generalizing from a Few Examples: A Survey on Few-shot
Learning," ACM Computing Surveys (CSUR), 53(3), 1–34., 2020.
[3] 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, 2013.
[4] W.-Z. Huang and S.-T. Cheng, "An Intelligence Evolution Platform for Intelligence Growth and Sharing," 2021.
[5] A. Farhadi, I. Endres, D. Hoiem, and D. Forsyth, "Describing objects by their attributes," 2009.
[6] C. H. Lampert, H. Nickisch, and S. Harmeling, "Learning to detect unseen object classes by between-class attribute transfer," 2009.
[7] J. Lu, J. Li, Z. Yan, and C. Zhang, "Zero-Shot Learning by Generating Pseudo Feature Representations," ArXiv, vol. abs/1703.06389, 2017.
[8] R. Gao et al., "Zero-VAE-GAN: Generating Unseen Features for Generalized and Transductive Zero-Shot Learning," IEEE Transactions on Image Processing ( Volume: 29), 2020.
[9] A. Frome et al., "DeViSE: A deep visual-semantic embedding model," 2013.
[10] K. He, X. Zhang, S. Ren, and J. Sun, "Deep Residual Learning for Image Recognition," Proceedings of the IEEE conference on computer vision and pattern recognition, 2016.
[11] T. Mikolov, K. Chen, G. Corrado, and J. Dean, "Efficient Estimation of Word Representations in Vector Space," p. arXiv:1301.3781. [Online]. Available: https://ui.adsabs.harvard.edu/abs/2013arXiv1301.3781M
[12] J. Pennington, R. Socher, and C. Manning, GloVe: Global Vectors for Word Representation. Doha, Qatar: Association for Computational Linguistics, 2014, pp. 1532-1543.
[13] A. Joulin, E. Grave, P. Bojanowski, and T. Mikolov, "Bag of Tricks for Efficient Text Classification," p. arXiv:1607.01759. [Online]. Available: https://ui.adsabs.harvard.edu/abs/2016arXiv160701759J
[14] D. Wolpert, "Stacked Generalization.," Neural Networks, 1992/12/31.
[15] Sam T. Roweis and Lawrence K. Saul, "Nonlinear Dimensionality Reduction by Locally Linear Embedding," 2000.
[16] Y. Xian, C. H. Lampert, B. Schiele, and Z. Akata, "Zero-Shot Learning—A Comprehensive Evaluation of the Good, the Bad and the Ugly," TPAMI, 2019.
[17] E. Kodirov, T. Xiang, and S. Gong, "Semantic Autoencoder for Zero-Shot Learning," p. arXiv:1704.08345. [Online]. Available: https://ui.adsabs.harvard.edu/abs/2017arXiv170408345K