近幾年來，僅利用少量標籤資料來訓練模型並分類新類別已備受廣大注意，此類的任務稱為少樣本分類學習。雖然在此領域中已有重大的進展,但多數現行少樣本分類學習之方法以目標與來源資料來自相同的分布為前提。因此大多數的少樣本分類學習方法在目標領域的一般化效果都不盡理想。本篇論文，我們解決上述少樣本分類於領域相異之情境。更重要的是，我們提出的方法不僅解決少樣本分類於領域相異之情境，特別在來源資料所包含概念較不充分的情況，此情境我們稱為「跨領域少樣本學習自概念貧乏之來源資料」。我們處理的問題是，模型訓練自概念貧乏的來源資料，且需要一般化到目標領域。在此困境下，我們提出了簡單有效的dropout風格的方法來幫助已訓練之模型更好的一般化到目標領域。我們的主要想法是利用類似 dropout 的機制來產生多個特徵多樣化之模型，並從中篩選出對目標資料適應較好的幾個模型來集成一個強的模型。我們在此情境設定下的實驗結果顯示，我們提出的方法能夠套用於各個不同的少樣本學習方法，讓模型學到更豐富的概念，並且都能改善至更高的準確度。

The goal of few-shot classification is to classify unseen categories with few labeled data. This area has attracted considerable attention. While significant progress has been made, most of existing few-shot classification methods assumed target data and source data came from the same distribution. Therefore, these methods often failed to generalize to unseen domains. In this work, the problem of few-shot classification with domain shift is tackled. Furthermore, we focus on a more challenging problem. We propose a novel scenario called \emph{cross-domain few-shot learning from naive concept}. We deal with the problem that the model is trained with source data with insufficient concept and should generalize to target domain. Then, we propose a simple yet effective dropout-style method for a trained model to generalize better to the target domain. The main idea is to sample several sub-networks by dropping neurons (or feature maps) to construct a bunch of models with diverse features for target domain. Then we choose those suitable for target domain to do the ensemble. We conduct experiments under the domain shift from naive concept setting and shows that the proposed method is applicable to various metric-based models, and improves consistently on the few-shot classification with domain shift from naive concept.

[1] L. Fei-Fei, R. Fergus, and P. Perona, “One-shot learning of object categories,” IEEE transactions on
pattern analysis and machine intelligence, vol. 28, no. 4, pp. 594–611, 2006.
[2] S. J. Pan and Q. Yang, “A survey on transfer learning,” IEEE Transactions on knowledge and data
engineering, vol. 22, no. 10, pp. 1345–1359, 2009.
[3] O. Vinyals, C. Blundell, T. Lillicrap, K. Kavukcuoglu, and D. Wierstra, “Matching networks for one
shot learning. nips (2016).”
[4] J. Snell, K. Swersky, and R. Zemel, “Prototypical networks for few-shot learning,” in Advances in
neural information processing systems, pp. 4077–4087, 2017.
[5] F. Sung, Y. Yang, L. Zhang, T. Xiang, P. H. Torr, and T. M. Hospedales, “Learning to compare: Relation
network for few-shot learning,” in Proceedings of the IEEE Conference on Computer Vision and
Pattern Recognition, pp. 1199–1208, 2018.
[6] C. Finn, P. Abbeel, and S. Levine, “Model-agnostic meta-learning for fast adaptation of deep networks,”
arXiv preprint arXiv:1703.03400, 2017.
[7] Z. Li, F. Zhou, F. Chen, and H. Li, “Meta-sgd: Learning to learn quickly for few-shot learning,” arXiv
preprint arXiv:1707.09835, 2017.
[8] A. Antoniou, H. Edwards, and A. Storkey, “How to train your maml,” arXiv preprint arXiv:
1810.09502, 2018.
[9] S. Ravi and H. Larochelle, “Optimization as a model for few-shot learning,” 2016.
[10] Y. Guo, N. C. Codella, L. Karlinsky, J. R. Smith, T. Rosing, and R. Feris, “A new benchmark for
evaluation of cross-domain few-shot learning,” arXiv preprint arXiv:1912.07200, 2019.
[11] W.-Y. Chen, Y.-C. Liu, Z. Kira, Y.-C. F. Wang, and J.-B. Huang, “A closer look at few-shot classification,”
arXiv preprint arXiv:1904.04232, 2019.
[12] X. Chen, H. Dai, Y. Li, X. Gao, and L. Song, “Learning to stop while learning to predict,” arXiv
preprint arXiv:2006.05082, 2020.
[13] A. Santoro, S. Bartunov, M. Botvinick, D. Wierstra, and T. Lillicrap, “Meta-learning with memoryaugmented
neural networks,” in International conference on machine learning, pp. 1842–1850, 2016.
[14] T. Munkhdalai and H. Yu, “Meta networks,” in Proceedings of the 34th International Conference on
Machine Learning-Volume 70, pp. 2554–2563, 2017.
[15] T. Ramalho and M. Garnelo, “Adaptive posterior learning: few-shot learning with a surprise-based
memory module,” in International Conference on Learning Representations, 2018.
35
[16] B. Hariharan and R. Girshick, “Low-shot visual recognition by shrinking and hallucinating features,”
in Proceedings of the IEEE International Conference on Computer Vision, pp. 3018–3027, 2017.
[17] Y.-X. Wang, R. Girshick, M. Hebert, and B. Hariharan, “Low-shot learning from imaginary data,”
in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 7278–7286,
2018.
[18] H. Zhang, J. Zhang, and P. Koniusz, “Few-shot learning via saliency-guided hallucination of samples,”
in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2770–2779,
2019.
[19] A. Alfassy, L. Karlinsky, A. Aides, J. Shtok, S. Harary, R. Feris, R. Giryes, and A. M. Bronstein,
“Laso: Label-set operations networks for multi-label few-shot learning,” in Proceedings of the IEEE
Conference on Computer Vision and Pattern Recognition, pp. 6548–6557, 2019.
[20] W.-H. Chu, Y.-J. Li, J.-C. Chang, and Y.-C. F. Wang, “Spot and learn: A maximum-entropy patch sampler
for few-shot image classification,” in Proceedings of the IEEE Conference on Computer Vision
and Pattern Recognition, pp. 6251–6260, 2019.
[21] Z. Chen, Y. Fu, Y.-X. Wang, L. Ma, W. Liu, and M. Hebert, “Image deformation meta-networks for
one-shot learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,
pp. 8680–8689, 2019.
[22] V. G. Satorras and J. B. Estrach, “Few-shot learning with graph neural networks,” in International
Conference on Learning Representations, 2018.
[23] B. Oreshkin, P. R. López, and A. Lacoste, “Tadam: Task dependent adaptive metric for improved
few-shot learning,” in Advances in Neural Information Processing Systems, pp. 721–731, 2018.
[24] L. Bertinetto, J. F. Henriques, P. Torr, and A. Vedaldi, “Meta-learning with differentiable closed-form
solvers,” in International Conference on Learning Representations, 2018.
[25] Y. Lifchitz, Y. Avrithis, S. Picard, and A. Bursuc, “Dense classification and implanting for fewshot
learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,
pp. 9258–9267, 2019.
[26] I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and
Y. Bengio, “Generative adversarial nets,” in Advances in neural information processing systems,
pp. 2672–2680, 2014.
[27] Y. Ganin and V. Lempitsky, “Unsupervised domain adaptation by backpropagation,” in International
conference on machine learning, pp. 1180–1189, 2015.
[28] N. Dong and E. P. Xing, “Domain adaption in one-shot learning,” in Joint European Conference on
Machine Learning and Knowledge Discovery in Databases, pp. 573–588, Springer, 2018.
36
[29] N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, “Dropout: a simple way
to prevent neural networks from overfitting,” The journal of machine learning research, vol. 15, no. 1,
pp. 1929–1958, 2014.
[30] G. E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever, and R. R. Salakhutdinov, “Improving neural
networks by preventing co-adaptation of feature detectors,” arXiv preprint arXiv:1207.0580, 2012.
[31] J. Tompson, R. Goroshin, A. Jain, Y. LeCun, and C. Bregler, “Efficient object localization using convolutional
networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition,
pp. 648–656, 2015.
[32] M. Stone, “Cross-validatory choice and assessment of statistical predictions,” Journal of the Royal
Statistical Society: Series B (Methodological), vol. 36, no. 2, pp. 111–133, 1974.
[33] B. Lake, R. Salakhutdinov, J. Gross, and J. Tenenbaum, “One shot learning of simple visual concepts,”
in Proceedings of the annual meeting of the cognitive science society, vol. 33, 2011.
[34] G. Cohen, S. Afshar, J. Tapson, and A. van Schaik, “Emnist: an extension of mnist to handwritten
letters. arxiv e-prints,” arXiv preprint arXiv:1702.05373, 2017.
[35] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “ImageNet: A Large-Scale Hierarchical
Image Database,” in CVPR09, 2009.
[36] C. Wah, S. Branson, P. Welinder, P. Perona, and S. Belongie, “The caltech-ucsd birds-200-2011 dataset
(2011),” California Institute of Technology.
[37] 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, pp. 770–778, 2016.