Page 154 - 《软件学报》2020年第12期

3820 Journal of Software 软件学报 Vol.31, No.12, December 2020

Table 4 The architecture of the convolutional network of Mean Teacher model
表 4 Mean Teacher 模型的网络结构和参数
网络层 超参数
输入 28(32)×28(32)像素图像
转化(translation) 随机{Δx,Δy}~[−2,2]
水平反转(horizontal flip) 随机 p=0.5
高斯噪声 σ=0.15
卷积层 128 卷积核 3×3,填充为 0(same padding)
卷积层 128 卷积核 3×3,填充为 0(same padding)
卷积层 128 卷积核 3×3,填充为 0(same padding)
池化层 最大池化 2×2
Dropout p=0.5
卷积层 256 卷积核 3×3,填充为 0(same padding)
卷积层 256 卷积核 3×3,填充为 0(same padding)
卷积层 256 卷积核 3×3,填充为 0(same padding)
池化层 最大池化 2×2
Dropout p=0.5
卷积层 512 卷积核 3×3,“丢弃”(valid padding)
卷积层 256 卷积核 1×1,填充为 0(same padding)
卷积层 128 卷积核 1×1,填充为 0(same padding)
池化层 平均池化(6×6→1×1)
softmax 全连接

References:
[1] Andriluka M, Pishchulin L, Gehler P, Schiele B. 2D human pose estimation: New benchmark and state of the art analysis. In: Proc.
of the IEEE Conf. on Computer Vision and Pattern Recognition. 2014. 3686−3693.
[2] Settles B. Active learning literature survey. Technical Report 1648. University of Wisconsin-Madison, 2009.
[3] Tong S, Koller D. Support vector machine active learning with applications to text classification. Journal of Machine Learning
Research, 2001,2(Nov.):45−66.
[4] Xu Z, Yu K, Tresp V, Xu X, Wang J. Representative sampling for text classification using support vector machines. In: Proc. of the
European Conf. on Information Retrieval. Berlin, Heidelberg: Springer-Verlag, 2003. 393−407.
[5] Brinker K. Incorporating diversity in active learning with support vector machines. In: Proc. of the Int’l Conf. on Machine Learning.
2003. 59−66.
[6] Chattopadhyay R, Wang Z, Fan W, Davidson I, Panchanathan S, Ye J. Batch mode active sampling based on marginal probability
distribution matching. ACM Trans. on Knowledge Discovery from Data, 2013,7(3):1−25.
[7] Wang Z, Ye J. Querying discriminative and representative samples for batch mode active learning. ACM Trans. on Knowledge
Discovery from Data, 2015,9(3):1−23.
[8] Zhang Y, Lease M, Wallace BC. Active discriminative text representation learning. In: Proc. of the AAAI Conf. on Artificial
Intelligence. 2017. 3386−3392.
[9] Donmez P, Carbonell JG, Bennett PN. Dual strategy active learning. In: Proc. of the European Conf. on Machine Learning. Berlin,
Heidelberg: Springer-Verlag, 2007. 116−127.
[10] Huang SJ, Jin R, Zhou ZH. Active learning by querying informative and representative examples. In: Proc. of the Advances in
Neural Information Processing Systems. 2010. 892−900.
[11] Du B, Wang Z, Zhang L, Zhang L, Liu W, Shen J, Tao D. Exploring representativeness and informativeness for active learning.
IEEE Trans. on Cybernetics, 2017,47(1):14−26.
[12] Li YC, Wang YL, Yu DJ, Hu P, Zhao RX. ASCENT: Active supervision for Semi-Supervised learning. IEEE Trans. on Knowledge
and Data Engineering, 2020,32(5):868−882.
[13] Hoi SC, Jin R, Zhu J, Lyu MR. Semi-Supervised SVM batch mode active learning with applications to image retrieval. ACM Trans.
on Information Systems, 2009,27(3):1−29.
[14] Yin C, Qian B, Cao S, Li X, Wei J, Zheng Q, Davidson I. Deep similarity-based batch mode active learning with exploration-
exploitation. In: Proc. of the IEEE Int’l Conf. on Data Mining. 2017. 575−584.