基于Kullback-Leiber距離的遷移仿射聚類算法
doi: 10.11999/JEIT151132
基金項(xiàng)目:
國家自然科學(xué)基金(61170122, 61272210)，江蘇省 2014 年度普通高校研究生科研創(chuàng)新計(jì)劃項(xiàng)目(KYLX_1124)，山東省高等學(xué)校科技計(jì)劃項(xiàng)目(J14LN05)
Transfer Affinity Propagation Clustering Algorithm Based on Kullback-Leiber Distance
Funds:
The National Natural Science Foundation of China (61170122, 71272210), Jiangsu Graduate Student Innovation Projects (KYLX_1124), The Science and Technology Program Shandong Provinceial Higher Education (J14LN05)
-
摘要: 針對遷移聚類問題，該文提出一種新的基于Kullback-Leiber距離的遷移仿射聚類算法(TAP_KL)。該算法從概率角度重新解釋AP算法的目標(biāo)函數(shù)，并借助于信息論中最常見的一種距離度量，即Kullback-Leiber距離，測量源域與目標(biāo)域代表點(diǎn)的相似性。另外，通過詳細(xì)分析TAP_KL算法與AP算法的目標(biāo)函數(shù)，得出一個(gè)重要結(jié)論，即可以將源域與目標(biāo)域的相似性嵌入到目標(biāo)域數(shù)據(jù)集相似性矩陣的計(jì)算中，從而直接利用AP算法的優(yōu)化算法優(yōu)化TAP_KL算法的目標(biāo)函數(shù)，解決基于代表點(diǎn)的遷移聚類問題。最后，通過基于4個(gè)數(shù)據(jù)集的仿真實(shí)驗(yàn)，進(jìn)一步驗(yàn)證了TAP_KL算法在解決遷移聚類問題時(shí)的有效性。
-
關(guān)鍵詞:
- 仿射聚類算法 /
- 遷移學(xué)習(xí) /
- 人臉數(shù)據(jù)集 /
- 概率框架 /
- KL距離
Abstract: For solving the clustering problem of transfer learning, a new algorithm called Transfer Affinity Propagation clustering algorithm is proposed based on Kullback-Leiber distance (TAP_KL). Based on the probabilistic framework, a new interpretation of the objective function of Affinity Propagation (AP) clustering algorithm is proposed. By leveraging Kullback-Leiber distance which is usually used in information theory, TAP_KL measures the similarity relationship between source data and target data. Moreover, TAP_KL algorithm can embed the similarity relationship to the calculation of similarity matrix of target data. Thus, the optimization framework of AP can be directly used to optimize the new target function of TAP_KL. In this case, TAP_KL builds a simple algorithm framework to solve the transfer clustering problem, in which the algorithm just needs to modify the similarity matrix to solve the transfer clustering problem. The experimental results based on both 4 datasets show the effectiveness of the proposed algorithm TAP_KL. -
LONG M, WANG J, DING G, et al. Adaptation regularization: A general framework for transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2014, 26(5): 1076-1089. doi: 10.1109/TKDE.2013.111. 畢安琪, 王士同. 基于SVC和SVR約束組合的遷移學(xué)習(xí)分類算法[J]. 控制與決策, 2014, 29(6): 1021-1026. doi: 10.13195 /j.kzyjc.2013.0520. BI Anqi and WANG Shitong. Transfer classification learning based on combination of both SVC and SVRs constraints[J]. Control and Decision, 2014, 29(6): 1021-1026. doi: 10.13195/j. kzyjc.2013.0520. PATRICIA N and CAPUTO B. Learning to learn, from transfer learning to domain adaptation: A unifying perspective[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 2014: 1442-1449. doi: 10.1109/CVPR.2014.187. XU Z J and SUN S L. Multi-source transfer learning with multi-view adaboost[J]. Neural Information Processing, 2012, 7665: 332-339. doi: 10.1007/978-3-642-34487-9_41. PAN S J L, KWOK J T, and YANG Q. Transfer learning via dimensionality reduction[C]. Proceedings of the 23rd International Conference on Artificial Intelligence, CA, USA: 2008: 677-682. PAN S J L, NI X C, SUN J T, et al. Cross domain sentiment classification via spectral feature alignment[C]. Proceedings of the 19th International Conference on World Wide Web (WWW-10). New York, USA, 2010, 751-760. doi: 10. 1145/1772690.1772767. 蔣亦樟, 鄧趙紅, 王士同. ML型遷移學(xué)習(xí)模糊系統(tǒng)[J]. 自動(dòng)化學(xué)報(bào), 2012, 38(9): 1393-1409. doi: 10.3724/SP.J.1004.2012. 01393. JIANG Yizhang, DENG Zhaohong, and WANG Shitong. Mamdani-Larsen type transfer learning fuzzy system[J]. Acta Automatica Sinica, 2012, 38(9): 1393-1409. doi: 10. 3724/SP.J.1004.2012.01393. JIANG W H and CHUNG F L. Transfer spectral clustering [C]. Proceedings of the 2012 European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, Berlin, Heidelberg, 2012: 789-803. doi: 10.1007/978-3-642-33486-3_50. LI M J, NG M K, CHEUNG Y M, et al. Agglomerative fuzzy K-means clustering algorithm with selection of number of clusters[J]. IEEE Transactions on Knowledge and Data Engineering, 2008, 20(11): 1519-1534. doi: 10.1109/TKDE. 2008.88. KRISHMA K and MURTY M N. Genetic K-means algorithm[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part B, 1999, 29(3): 433-439. doi: 10.1109/3477. 764879. ERSAHIN K, CUMMING I G, and WARD R K. Segmentation and classification of Polari metric SAR data using spectral graph partitioning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2010, 48(1): 164-167. doi: 10.1109/TGRS.2009.2024303. LAUER F and SCHNORR C. Spectral clustering of linear subspaces for motion segmentation[C]. Proceedings of the 12th IEEE International Conference of Computer Vision, Kyoto, Japan, 2009: 678-685. doi: 10.1109/ICCV. 2009. 5459173. FREY B J and DUECK D. Clustering by passing messages between data points[J]. Science, 2007, 315(5814): 972-976. 肖宇, 于劍. 基于近鄰傳播算法的半監(jiān)督聚類[J]. 軟件學(xué)報(bào), 2008, 19(11): 2803-2813. doi: 10.3724/SP.J.1001.2008.02803. XIAO Yu and YU Jian. Semi-supervised clustering based on affinity propagation algorithm[J]. Journal of Software, 2008, 19(11): 2803-2813. doi: 10.3724/SP.J.1001.2008.02803. 儲(chǔ)岳中, 徐波, 高有濤. 基于近鄰傳播聚類與核匹配追蹤的遙感圖像目標(biāo)識(shí)別方法[J]. 電子與信息學(xué)報(bào), 2014, 36(12): 2923-2928. doi: 10.3724/SP.J.1146.2014.00422. CHU Yuezhong, XU Bo, and GAO Youtao. Technique of remote sensing image target recognition based on affinity propagation and kernel matching pursuit[J]. Journal of Electronics Information Technology, 2014, 36(12): 2923-2928. doi: 10.3724/SP.J.1146.2014.00422. SUN L and GUO C H. Incremental affinity propagation clustering based on message passing[J]. IEEE Transactions on Knowledge and Data Engineering, 2014, 26(11): 2731-2744. doi: 10.1109/TKDE.2014.2310215. SHI X H, GUAN R C, WANG L P, et al. An incremental affinity propagation algorithm and its applications for text clustering[C]. Proceedings International Joint Conference on Neural Networks, Atlanta, GA, USA, 2009: 2914-2919. doi: 10.1109/IJCNN.2009.5178973. ZHENG Yun and CHEN Pei. Clustering based on enhanced -expansion move[J]. IEEE Transactions on Knowledge and Data Engineering, 2013, 25(10): 2206-2216. doi: 10.1109/ TKDE.2012.202. 畢安琪, 董愛美, 王士同. 基于概率和代表點(diǎn)的數(shù)據(jù)流動(dòng)態(tài)聚類算法[J]. 計(jì)算機(jī)研究與發(fā)展, 2016, 53(5): 1029-1042. BI Anqi, DONG Aimei, and WANG Shitong. A dynamic data stream clustering algorithm based on probability and exemplar[J]. Journal of Computer Research and Development, 2016, 53(5): 1029-1042. 孫力娟, 陳小東, 韓崇, 等. 一種新的數(shù)據(jù)流模糊聚類方法[J]. 電子與信息學(xué)報(bào), 2015, 37(7): 1620-1625. doi: 10.11999/ JEIT141415. SUN Lijuan, CHEN Xiaodong, HAN Chong, et al. New fuzzy- clustering algorithm for data stream[J]. Journal of Electronics Information Technology, 2015, 37(7): 1620-1625. doi: 10.11999/JEIT141415. JIANG Y Z, CHUNG F L, WANG S T, et al. Collaborative fuzzy clustering from multiple weighted views[J]. IEEE Transactions on Cybernetics, 2015, 45(4): 688-701. doi: 10. 1109/TCYB.2014.2334595. CAI D, HE X F, HAN J W, et al. Orthogonal Laplacian faces for face recognition[J]. IEEE Transactions on Image Processing, 2006, 15(11): 3608-3614. doi: 10.1109/TIP.2006. 881945. 張景祥, 王士同, 鄧趙紅, 等. 融合異構(gòu)特征的子空間遷移學(xué)習(xí)算法[J]. 自動(dòng)化學(xué)報(bào), 2014, 40(2): 236-246. doi: 10.3724/SP.J. 1004.2014.00236. ZHANG Jingxiang, WANG Shitong, DENG Zhaohong, et al. A subspace transfer learning algorithm integrating heterogeneous features[J]. Acta Automatica Sinica, 2014, 40(2): 236-246. doi: 10.3724/SP.J.1004.2014.00236. LE C Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324. doi: 10.1109/5.726791. REN J, SHI X, FAN W, et al. Type independent correction of sample selection bias via structural discovery and re- balancing[C]. Proceedings of the 8th SIAM International Conference on Data Mining, Atlanta, GA, USA, 2008: 565-576. doi: 10.1137/1.9781611972788.52. -
計(jì)量
- 文章訪問數(shù): 1217
- HTML全文瀏覽量: 137
- PDF下載量: 450
- 被引次數(shù): 0
下載:
