Abstract
In order to obtain prominent recognition accuracy convolutional neural networks (CNNs) need large amounts of labeled data to initialize network parameters. However, there exist two open problems, i.e., the uncertainties of the initialized effects and the limited labeled data To address the problems, we propose a novel method named UPSCNNs, which uses unlabeled data to perform Principal Component Analysis (PCA) and shuffling initialization for CNNs, composed of four steps, i.e. sampling the input images, calculating the sampling sets with PCA and initializing and shuffling the convolutional kernels. In cases with the same network architecture and activation function, i.e., Rectified Linear Units, we conduct the comparative experiments on three image datasets, i.e., STL-10, CIFAR-10(I) and CIFAR-10(II). In terms of accuracy, we find (1) the novel method increases by 4-20 percent in comparison to other weight initialization methods, e.g., Msra initialization, Xavier initialization and Random initialization and (2) an increase of 1-3 percent is obtained with unlabeled data than with only labeled data The results indicate that our method can make full use of unlabeled data for initializing CNNs to achieve good recognition effectiveness.
Similar content being viewed by others
Change history
25 August 2018
The original version of this article unfortunately contained a mistake. Errors in the online version as follows: (1) In the 2nd row of Column 8 (Rel(%)) in Table 5, the number “54814” should be changed to “5.4814”.
25 August 2018
The original version of this article unfortunately contained a mistake. Errors in the online version as follows: (1) In the 2nd row of Column 8 (Rel(%)) in Table 5, the number ?54814? should be changed to ?5.4814?.
References
Krizhevsky A, Sutskever I, Hinton GE (2012) Imagenet classification with deep convolutional neural networks. In: Advances in neural information processing systems, pp 1097–1105
Farabet C, Couprie C, Najman L, LeCun Y (2013) Learning hierarchical features for scene labeling. IEEE Trans Pattern Anal Mach Intell 35(8):1915–1929
Hinton G, Deng L, Yu D, Dahl GE, Mohamed AR, Jaitly N, Kingsbury B (2012) Deep neural networks for acoustic modeling in speech recognition: the shared views of four research groups. IEEE Signal Process Mag 29(6):82–97
Collobert R, Weston J, Bottou L, Karlen M, Kavukcuoglu K, Kuksa P (2011) Natural language processing (almost) from scratch. J Mach Learn Res 12:2493–2537
Mikolov T, Deoras A, Povey D, Burget L, Černocký J (2011) Strategies for training large scale neural network language models. In: 2011 IEEE workshop on automatic speech recognition and understanding (ASRU). IEEE, pp 196–201
Bordes A, Chopra S (2014) Question answering with subgraph embeddings. arXiv:1406.3676
Jean S, Cho K, Memisevic R, Bengio Y (2014) On using very large target vocabulary for neural machine translation. arXiv:1412.2007
Thimm G, Fiesler E (1995) Neural network initialization. In: International workshop on artificial neural networks. Springer, Berlin, pp 535–542
Glorot X, Bengio Y (2010) Understanding the difficulty of training deep feedforward neural networks. In: Proceedings of the 13th international conference on artificial intelligence and statistics, pp 249–256
Bengio Y (2012) Practical recommendations for gradient-based training of deep architectures. In: Neural networks: tricks of the trade. Springer, Berlin, pp 437–478
He K, Zhang X, Ren S, Sun J (2015) Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In: Proceedings of the IEEE international conference on computer vision, pp 1026–1034
Gu S, Jin Y (2017) Multi-train: a semi-supervised heterogeneous ensemble classifier. Neurocomputing 249:202–211
He G, Li Y, Zhao W (2017) An uncertainty and density based active semi-supervised learning scheme for positive unlabeled multivariate time series classification. Knowl-Based Syst 124:80–92
Karlos S, Fazakis N, Kotsiantis S, Sgarbas K (2017) Self-trained stacking model for semi-supervised learning. Int J Artif Intell Tools 26(2):1750001
Fazakis N, Karlos S, Kotsiantis S, Sgarbas K (2017) Self-trained rotation forest for semi-supervised learning. J Intell Fuzzy Syst 32(1):711–722
Grzeszick R, Fink GA (2016) An iterative partitioning-based method for semi-supervised annotation learning in image collections. Int J Pattern Recognit Artif Intell 30(2):1655005
Scalzo F, Hu X (2013) Semi-supervised detection of intracranial pressure alarms using waveform dynamics. Physiol Meas 34(4):465
Culp M, Michailidis G (2008) An iterative algorithm for extending learners to a semi-supervised setting. J Comput Graph Stat 17(3):545–571
Zhu J, Hoi SC, Lyu MR (2008) Face annotation using transductive kernel fisher discriminant. IEEE Trans Multimed 10(1):86– 96
Pfahringer B, Leschi C, Reutemann P (2007) Scaling up semi-supervised learning: an efficient and effective LLGC variant. In: Pacific-asia conference on knowledge discovery and data mining. Springer, Berlin, pp 236–247
Liu Y J, Lu S, Li D, Tong S (2017) Adaptive controller design-based ABLF for a class of nonlinear time-varying state constraint systems. IEEE Trans Syst Man Cybern Syst Hum 47(7):1546–1553
Li D P, Liu YJ, Tong S, Chen CP, Li D (2018) Neural networks-based adaptive control for nonlinear state constrained systems with input delay. IEEE Transactions on Cybernetics
Wang Y, Qiu Y, Thai T, Moore K, Liu H, Zheng B (2017) A two-step convolutional neural network based computer-aided detection scheme for automatically segmenting adipose tissue volume depicting on CT images. Comput Methods Programs Biomed 144:97–104
Yoon Y, Jeon HG, Yoo D, Lee JY, Kweon IS (2017) Light-field image super-resolution using convolutional neural network. IEEE Signal Process Lett 24(6):848–852
Fu X, Huang J, Ding X, Liao Y, Paisley J (2017) Clearing the skies: a deep network architecture for single-image rain removal. IEEE Trans Image Process 26(6):2944–2956
Ren S, He K, Girshick R, Sun J (2015) Faster r-cnn: Towards real-time object detection with region proposal networks. In: Advances in neural information processing systems, pp 91–99
Suárez-Paniagua V, Segura-Bedmar I, Martínez P (2017) Exploring convolutional neural networks for drug–drug interaction extraction. Database, 2017
Li C, Wang X, Liu W (2017) Neural features for pedestrian detection. Neurocomputing 238:420–432
Sui X, Zheng Y, Wei B, Bi H, Wu J, Pan X, Zhang S (2017) Choroid segmentation from optical coherence tomography with graph-edge weights learned from deep convolutional neural networks. Neurocomputing 237:332–341
Panda P, Sengupta A, Roy K (2017) Energy-efficient and improved image recognition with conditional deep learning. ACM J Emerg Technol Comput Syst (JETC) 13(3):33
Abdi H, Williams LJ (2010) Principal component analysis. Wiley Interdisciplinary Rev Comput Stat 2 (4):433–459
Krizhevsky A (2012) Cuda-convnet
Acharya UR, Fujita H, Oh SL, Raghavendra U, Tan JH, Adam M, Hagiwara Y (2018) Automated identification of shockable and non-shockable life-threatening ventricular arrhythmias using convolutional neural network. Futur Gener Comput Syst 79:952–959
Coates A, Ng A, Lee H (2011) An analysis of single-layer networks in unsupervised feature learning. In: Proceedings of the 4th international conference on artificial intelligence and statistics, pp 215–223
Hui KY (2013) Direct modeling of complex invariances for visual object features. In: International conference on machine learning, pp 352–360
Coates A, Ng AY (2011) Selecting receptive fields in deep networks. In: Advances in neural information processing systems, pp 2528–2536
Krizhevsky A, Hinton G (2009) Learning multiple layers of features from tiny images
Acknowledgments
This work was supported in part by the National Natural Science Foundation of China under Grant 61175004 and the Natural Science Foundation of Beijing Municipality under grant 4112009.
Author information
Authors and Affiliations
Corresponding author
Additional information
The original version of this article was revised: incorrect data in found in tables 5, 6 and 8, and figures 7, 9, 10 and 11 were corrected.
Rights and permissions
About this article
Cite this article
Ou, J., Li, Y. & Shen, C. Unlabeled PCA-shuffling initialization for convolutional neural networks. Appl Intell 48, 4565–4576 (2018). https://doi.org/10.1007/s10489-018-1230-2
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10489-018-1230-2