International Core Journal of Engineering 2020-26 | Page 125
2000 training sets and 500 test sets. We conducted a total of
five sets of tests, randomly extracting 1000 data from 2000
training sets as training data, and randomly extracting 100
test sets as test data. The method proposed in this paper is
compared with semi-supervised manifold regularization, and
the results obtained are shown in Fig. 2.
due to environmental influences in the face classification.
Experiments on the CASIA-FaceV5 dataset demonstrate that
this method has higher face classification accuracy than the
existing methods. The next step is to reduce its
computational cost.
R EFERENCES
[1]
[2]
[3]
[4]
[5]
Figure 2. Average testing accuracy on two methods.
It can be seen that the average accuracy of semi-
supervised manifold regularization for CASIA-FaceV5 face
classification is above 90%. The main reason is that CASIA-
FaceV5 is a relatively standard face data set with a small
amount of data. Compared with the semi-supervised
manifold regularization system, our semi-supervised
structured sparse graph data classification test accuracy has
almost better test accuracy, both exceeding and reaching
95%. On the other hand, through comparative tests, it can be
found that the semi-supervised structured sparse graph data
classification has a significant improvement in computational
cost compared to semi-supervised manifold regularization.
Therefore, it can be considered that the semi-supervised
structured sparse graph data classification has good
recognition accuracy and needs to be improved in terms of
efficiency.
[6]
[7]
[8]
[9]
[10]
[11]
V. C ONCULUSION
[12]
We propose a semi-supervised structured sparse graph
data classification method for face classification. The
manifold method of retaining the global data structure and
the local data structure is adopted, and the L2 norm is used to
limit the model space, and the complexity of the model is
controlled to alleviate the problem of large data difference
[13]
103
Kong X, Yu P S. Semi-supervised feature selection for graph
classification[C]濁澳 Proceedings of the 16th ACM SIGKDD
international conference on Knowledge discovery and data mining.
ACM, 2010: 793-802.
Kudo T, Maeda E, Matsumoto Y. An application of boosting to graph
classification[C]. Advances in neural information processing systems.
2005: 729-736.
Riesen K, Bunke H. Graph classification by means of Lipschitz
embedding[J]. IEEE Transactions on Systems, Man, and Cybernetics,
Part B (Cybernetics), 2009, 39(6): 1472-1483.
Camps-Valls G, Marsheva T V B, Zhou D. Semi-supervised graph-
based hyperspectral image classification[J]. IEEE Transactions on
Geoscience and Remote Sensing, 2007, 45(10): 3044-3054.
Zhou D, Hofmann T, Schölkopf B. Semi-supervised learning on
directed graphs[C]. Advances in neural information processing
systems. 2005: 1633-1640.
Luo R, Liu X, Wu Z. Feature Extraction of Hyperspectral Images
With Semi-supervised Sparse Graph Learning[C]. 2018 Fifth
International Workshop on Earth Observation and Remote Sensing
Applications (EORSA). IEEE, 2018: 1-4.
Li J, Rong Y, Cheng H, et al. Semi-Supervised Graph Classification:
A Hierarchical Graph Perspective[C]. The World Wide Web
Conference. ACM, 2019: 972-982.
Li L, Peng Y, Qiu G, et al. A survey of virtual sample generation
technology for face recognition[J]. Artificial Intelligence Review,
2018, 50(1): 1-20.
NASSIH B, AMINE A, NGADI M, et al. DCT and HOG Feature Sets
Combined with BPNN for Efficient Face Classification[J]. Procedia
Computer Science, 2019, 148: 116-125.
Gou J, Song J, Ou W, et al. Representation-based classification
methods with enhanced linear reconstruction measures for face
recognition[J]. Computers & Electrical Engineering, 2019, 79:
106451.
Zhu X, Goldberg A B. Introduction to semi-supervised learning[J].
Synthesis lectures on artificial intelligence and machine learning,
2009, 3(1): 1-130.
Fan M, Gu N, Qiao H, et al. Sparse regularization for semi-supervised
classification[J]. Pattern Recognition, 2011, 44(8): 1777-1784.
Tsang I W, Kwok J T. Large-scale sparsified manifold
regularization[C]. Advances in Neural Information Processing
Systems. 2007: 1401-1408.