Abstract:Local multiple kernel learning method could learn a specific combination kernel function for various samples according to the local space characteristics,therefore it has better discriminant ability.In this paper,we propose a local variable sparsity based multiple kernel learning method.In our method,the samples are divided into a few groups with a soft grouping method and the sparsity of kernel weights in various local spaces is determined by the similarity of kernels.We use an alternative optimization method to solve this problem.The experiment on synthetic dataset indicates that our method has a strong advantage in discriminative feature learning and against noise.Finally we apply our method into image scene classification and the accuracy is improved obviously.
[1] VAPNIK V.The Nature of Statistical Learning Theory[M].New York:Springer-Verlag,1995.
[2] 汪洪桥,孙富春,蔡艳宁,等.多核学习方法[J].自动化学报,2010,36(8):1037-1050. Wang H Q,Sun F C,Cai Y N,et al.On multiple kernel learning methods[J].Acta Automatica Sinica,2010,36(8):1037-1050.(in Chinese)
[3] Li X,Mao W,Jiang W.Multiple kernel learning based extreme learning machine for classification design[J].Neural Computing and Applications,2016,27(1):175-184.
[4] Wang C D,Lai J H,Yu P S.Multi-view clustering based on belief propagation[J].IEEE Transactions on Knowledge & Data Engineering,2016,28(4):1007-1021.
[5] Nazarpour A,Adibi P.Two-stage multiple kernel learning for supervised dimensionality reduction[J].Pattern Recognition,2015,48(5):1854-1862.
[6] Sahoo D,Hoi S C H,Li B.Online multiple kernel regression[A].Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining[C].USA:ACM,2014.293-302.
[7] 亓晓振,王庆.一种基于稀疏编码的多核学习图像分类方法[J].电子学报,2012,40(4):773-779. QI Xiao-zhen,WANG qing.An image classification approach based on sparse coding and multiple kernel learning[J].Acta Electronica Sinica,2012,40(4):773-779.(in Chinese)
[8] 余家林,孙季丰,李万益.基于多核稀疏编码的三维人体姿态估计[J].电子学报,2016,44(8):1899-1908. YU Jia-lin,SUN Ji-feng,LI Wan-yi.3D human pose estimation based on multi-kernel sparse coding[J].Acta Electronica Sinica,2016,44(8):1899-1908.(in Chinese)
[9] 苏娟,王百合,刘代志.一种基于拓扑约束的多核跟踪算法[J].电子学报,2015,43(2):353-357. SU Juan,WANG Bai-he,LIU Dai-zhi.A multi-kernel tracking algorithm based on topology constraint[J].Acta Electronica Sinica,2015,43(2):353-357.(in Chinese)
[10] 杨赛,赵春霞,刘凡.多核学习融合局部和全局特征的人脸识别算法[J].电子学报,2016,44(10):2344-2350. YANG Sai,ZHAO Chun-xia,LIU Fan.Fusion of local and global features using multiple kernel learning for face recognition[J].Acta Electronica Sinica,2016,44(10):2344-2350.(in Chinese)
[11] Mehment G,Ethem A.Multiple kernel learning algorithms[J].Journal of Machine Learning Research,2011,12:2211-2268.
[12] Lecun Y,Bengio Y,Hinton G.Deep learning[J].Nature,2015,521(7553):436-444.
[13] Lin Y-Y,Liu T-L,Fuh C-S.Local ensemble kernel learning for object category recognition[A].Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)[C].USA:IEEE,2007.1-8.
[14] Lin Y-Y,Tsai J-F,Liu T-L.Efficient discriminative local learning for object recognition[A].Proceedings of IEEE 12th International Conference on Computer Vision (ICCV)[C].USA:IEEE,2009.598-605.
[15] Gonen M,Alpaydin E.Localized algorithms for multiple kernel learning[J].Pattern Recognition,2013,46(3):795-807.
[16] Han Y,Yang K,Liu G.Lp-norm localized multiple kernel learning via semi-definite programming[J].IEEE Signal Processing Letters,2012,19(10):688-691.
[17] Yang J,Li Y,Tian Y,et al.Group-sensitive multiple kernel learning for object recognition[J].IEEE Transactions on Image Processing,2012,21(5):2838-2852.
[18] 张瑞杰,郭志刚,李弼程,等.基于E~2LSH-MKL的视觉语义概念检测[J].自动化学报,2012,38(10):1671-1678. Zhang Ruijie,Guo Zhigang,Li Bicheng,et al.A visual semantic concept detection algorithm based on E2LSH-MKL[J].Acta Automatica Sinica,2012,38(10):1671-1678.(in Chinese)
[19] Fu G Y,Wang Q C,Wang H Q,Bai D Y.Group based non-sparse localized multiple kernel learning algorithm for image classification[A].Proceedings of the 4th IEEE International Conference on Cloud Computing and Intelligence Systems[C].USA:IEEE,2016.191-195.
[20] Fu G Y,Wang Q C,Bai D Y,Li L L.Group based localized multiple kernel learning algorithm with lp-norm[J].International Journal of Innovative Computing,Information and Control,2016,6(12):1835-1849.
[21] Han Y,Yang K,Ma Y,et al.Localized multiple kernel learning via sample-wise alternating optimization[J].IEEE Transactions on Cybernetics,2013,44(1):127-148.
[22] Liu X,Wang L,Zhang J,et al.Sample-adaptive multiple kernel learning[A].Proceedings of the 28th AAAI Conference on Artificial Intelligence[C].USA:AAAI Press,2014.1-16.
[23] Lei Y,Binder A,Doganü,Kloft M.Theory and algorithms for the localized setting of learning kernels[J].The Journal of Machine Learning Research,2015,7(7):173-195.
[24] Antic B,Ommer B.Per-sample kernel adaptation for visual recognition and grouping[A].IEEE International Conference on Computer Vision[C].USA:IEEE,2015.1251-1259.
[25] Marius K,Ulf B,Soren S,Alexander Z.Lp-norm multiple kernel learning[J].The Journal of Machine Learning Research,2011,12(12):953-997.
[26] Cristianini N,Shawe-Taylor J,Elisseeff A,et al.On kernel-target alignment[A].Proceedings of the International Conference on Neural Information Processing Systems[C].USA:MIT Press,2001.367-373.
[27] Platt J C.Sequential minimal optimization:a fast algorithm for training support vector machines[A].Advances in Kernel Methods-Support Vector Learning[C].USA:Microsoft,1998.212-223.
[28] Chen L,Chen C L,Lu M.A multiple-kernel fuzzy C-means algorithm for image segmentation[J].IEEE Transactions on Systems Man & Cybernetics,2011,41(5):1263-1274.
[29] Shrivastava A,Patel V M,Chellappa R.Multiple kernel learning for sparse representation-based classification[J].IEEE Transactions on Image Processing,2014,23(7):3013-3024.
[30] Lazebnik S,Schmid C,Ponce J.Beyond bags of features:spatial pyramid matching for recognizing natural scene categories[A].Proceedings of IEEE Conference on Computer Vision & Pattern Recognition[C].USA:IEEE Computer Society,2006.2169-2178.
[31] Fei-Fei L,Fergus R,Perona P.Learning generative visual models from few training examples:an incremental bayesian approach tested on 101 object categories[J].Computer Vision & Image Understanding,2004,106(1):178.
[32] Ni B,Li T,Moulin P.Beta process multiple kernel learning[A].Computer Vision and Pattern Recognition[C].USA:IEEE,2014.963-970.