第6期 汤礼颖，等：一种卷积神经网络集成的多样性度量方法 ·1037· 出集成效果相对更好的模型组合，且集成模型的 [13]邢红杰，魏勇乐.基于相关嫡和距离方差的支持向量 性能对比基模型平均性能的提升效果也更优。此 数据描述选择性集成.计算机科学，2016,43(5)： 方法为基于概率向量输出的卷积神经网络模型集 252-256,264. 成选择提供了一种新的多样性度量思路。 XING Hongjie,WEI Yongle.Selective ensemble of SVDDs based on correntropy and distance variance[J]. 参考文献 Computer science,2016,43(5):252-256,264 [14]李莉.基于差异性度量的分类器集成优化方法研究与 [1]OPITZ D.MACLIN R.Popular ensemble methods:an 应用[D1.大连：大连海事大学，2017. empirical study[J].Journal of artificial intelligence re- LILi.Optimization method research and application of search,1999,11:169-198 multiple classifiers ensemble based on diversity meas- [2]ZHOU Zhuhui.Ensemble methods:foundations and al- ure[D].Dalian:Dalian Maritime University,2017 gorithms[M].New York:CRC Press,2012:236 [15]赵军阳，韩崇昭，韩德强，等.采用互补信息嫡的分类 [3]YULE G U.On the association of attributes in statistics: 器集成差异性度量方法).西安交通大学学报，2016， with illustrations from the material of the childhood soci- 50(2:13-19. ety,&c[J].Philosophical transactions of the royal society ZHAO Junyang,Han Chongzhao,Han Deqiang,et al.A of London..Series A,1900,1900,194:257-319 novel measure method for diversity of classifier integra- [4]SKALAK D B.The sources of increased accuracy for two tions using complement informationentropy[J].Journal proposed boosting algorithms[C]//Proceedings of Americ- of Xi'an Jiaotong University,2016,50(2):13-19 an Association for Artificial Intelligence,AAAI-96,In- [16]周钢，郭福亮.基于信息嫡的集成学习过程多样性度 tegrating Multiple Learned Models Workshop.Portland, 量研究).计算机工程与科学，2019,41(9)：1700 USA,1996:1133 1707. [5]GIACINTO G,ROLI F.Design of effective neural net- ZHOU Gang,GUO Fuliang.Process diversity measure- work ensembles for image classification purposes[J].Im- ment of ensemble learning based on information en age and vision computing,2001,19(9/10):699-707. tropy[J].Computer engineering and science,2019, [6]KUNCHEVA L I,WHITAKER C J.Measures of di- 41(9):1700-1707. versity in classifier ensembles and their relationship with [1刀周飞燕，金林鹏，董军.卷积神经网络研究综述.计 the ensemble accuracy[J].Machine learning,2003,51(2): 算机学报，2017,40(6)：1229-1251. 181-207 ZHOU Feiyan,JIN Linpeng,DONG Jun.DONG Jun [7]KOHAVI R.WOLPERT D H.Bias plus variance decom- Review of convolutional neural network[J].Chinese position for zero-one loss functions[C]//Proceedings of journal of computers,2017,40(6):1229-1251. the 13th International Conference on Machine Learning. [18]FAN Tiegang.ZHU Ying,CHEN Junmin.A new meas- San Francisco,USA,1996:275-283. ure of classifier diversity in multiple classifier system [8]CONOVER W J.Statistical methods for rates and propor- [C]//Proceedings of 2008 International Conference on tions[J].Technometrics,1974,16(2):326-327 Machine Learning and Cybernetics.Kunming,China, [9]SHIPP C A.KUNCHEVA L I.Relationships between 2008. combination methods and measures of diversity in com- [19]常亮，邓小明，周明全，等.图像理解中的卷积神经网 bining classifiers[J].Information fusion,2002,3(2): 络[J.自动化学报，2016,42(9)：1300-1312 135-148 CHANG Liang,DENG Xiaoming,ZHOU Mingquan,et [10]HANSEN L K,SALAMON P.Neural network en- al.Convolutional neural networks in image understand- sembles[J].IEEE transactions on pattern analysis and ing[J].Acta Automatica Sinica,2016,42(9):1300-1312 machine intelligence,2002,12(10):993-1001 [20]KRIZHEVSKY A,HINTON G.Learning multiple lay- [11]CUNNINGHAM P,CARNEY J.Diversity versus qual- ers of features from tiny images[J].Handbook of system- ity in classification ensembles based on feature selec- ic autoimmune diseases,2009.1(4):7. tion[Cl//Proceedings of the 11th European Conference [21]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.Im- on Machine Learning.Catalonia,Spain,2000:109-116. ageNet classification with deep convolutional neural net- [12]PARTRIDGE D,KRZANOWSKI W.Software di- works[C]//Proceedings of the 25th International Confer- versity:practical statistics for its measurement and ex- ence on Neural Information Processing Systems.Lake ploitation[J].Information and software technology, Tahoe.USA.2012. 1997,3910):707-717. [22]SINHA N K.GRISCIK M P.A stochastic approxima-出集成效果相对更好的模型组合,且集成模型的 性能对比基模型平均性能的提升效果也更优。此 方法为基于概率向量输出的卷积神经网络模型集 成选择提供了一种新的多样性度量思路。 参考文献： OPITZ D, MACLIN R. Popular ensemble methods: an empirical study[J]. Journal of artificial intelligence re￾search, 1999, 11: 169–198. [1] ZHOU Zhuhui. Ensemble methods: foundations and al￾gorithms[M]. New York: CRC Press, 2012: 236. [2] YULE G U. On the association of attributes in statistics: with illustrations from the material of the childhood soci￾ety, &c[J]. Philosophical transactions of the royal society of London. Series A, 1900, 1900, 194: 257–319. [3] SKALAK D B. The sources of increased accuracy for two proposed boosting algorithms[C]//Proceedings of Americ￾an Association for Artificial Intelligence, AAAI-96, In￾tegrating Multiple Learned Models Workshop. Portland, USA, 1996: 1133. [4] GIACINTO G, ROLI F. Design of effective neural net￾work ensembles for image classification purposes[J]. Im￾age and vision computing, 2001, 19(9/10): 699–707. [5] KUNCHEVA L I, WHITAKER C J. Measures of di￾versity in classifier ensembles and their relationship with the ensemble accuracy[J]. Machine learning, 2003, 51(2): 181–207. [6] KOHAVI R, WOLPERT D H. Bias plus variance decom￾position for zero-one loss functions[C]//Proceedings of the 13th International Conference on Machine Learning. San Francisco, USA, 1996: 275−283. [7] CONOVER W J. Statistical methods for rates and propor￾tions[J]. Technometrics, 1974, 16(2): 326–327. [8] SHIPP C A, KUNCHEVA L I. Relationships between combination methods and measures of diversity in com￾bining classifiers[J]. Information fusion, 2002, 3(2): 135–148. [9] HANSEN L K, SALAMON P. Neural network en￾sembles[J]. IEEE transactions on pattern analysis and machine intelligence, 2002, 12(10): 993–1001. [10] CUNNINGHAM P, CARNEY J. Diversity versus qual￾ity in classification ensembles based on feature selec￾tion[C]//Proceedings of the 11th European Conference on Machine Learning. Catalonia, Spain, 2000: 109−116. [11] PARTRIDGE D, KRZANOWSKI W. Software di￾versity: practical statistics for its measurement and ex￾ploitation[J]. Information and software technology, 1997, 39(10): 707–717. [12] 邢红杰, 魏勇乐. 基于相关熵和距离方差的支持向量 数据描述选择性集成 [J]. 计算机科学, 2016, 43(5): 252–256, 264. XING Hongjie, WEI Yongle. Selective ensemble of SVDDs based on correntropy and distance variance[J]. Computer science, 2016, 43(5): 252–256, 264. [13] 李莉. 基于差异性度量的分类器集成优化方法研究与 应用 [D]. 大连: 大连海事大学, 2017. LI Li. Optimization method research and application of multiple classifiers ensemble based on diversity meas￾ure[D]. Dalian: Dalian Maritime University, 2017. [14] 赵军阳, 韩崇昭, 韩德强, 等. 采用互补信息熵的分类 器集成差异性度量方法 [J]. 西安交通大学学报, 2016, 50(2): 13–19. ZHAO Junyang, Han Chongzhao, Han Deqiang, et al. A novel measure method for diversity of classifier integra￾tions using complement informationentropy[J]. Journal of Xi’an Jiaotong University, 2016, 50(2): 13–19. [15] 周钢, 郭福亮. 基于信息熵的集成学习过程多样性度 量研究 [J]. 计算机工程与科学, 2019, 41(9): 1700– 1707. ZHOU Gang, GUO Fuliang. Process diversity measure￾ment of ensemble learning based on information en￾tropy[J]. Computer engineering and science, 2019, 41(9): 1700–1707. [16] 周飞燕, 金林鹏, 董军. 卷积神经网络研究综述 [J]. 计 算机学报, 2017, 40(6): 1229–1251. ZHOU Feiyan, JIN Linpeng, DONG Jun. DONG Jun. Review of convolutional neural network[J]. Chinese journal of computers, 2017, 40(6): 1229–1251. [17] FAN Tiegang, ZHU Ying, CHEN Junmin. A new meas￾ure of classifier diversity in multiple classifier system [C]//Proceedings of 2008 International Conference on Machine Learning and Cybernetics. Kunming, China, 2008. [18] 常亮, 邓小明, 周明全, 等. 图像理解中的卷积神经网 络 [J]. 自动化学报, 2016, 42(9): 1300–1312. CHANG Liang, DENG Xiaoming, ZHOU Mingquan, et al. Convolutional neural networks in image understand￾ing[J]. Acta Automatica Sinica, 2016, 42(9): 1300–1312. [19] KRIZHEVSKY A, HINTON G. Learning multiple lay￾ers of features from tiny images[J]. Handbook of system￾ic autoimmune diseases, 2009, 1(4): 7. [20] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Im￾ageNet classification with deep convolutional neural net￾works[C]//Proceedings of the 25th International Confer￾ence on Neural Information Processing Systems. Lake Tahoe, USA, 2012. [21] [22] SINHA N K, GRISCIK M P. A stochastic approxima- 第 6 期 汤礼颖，等：一种卷积神经网络集成的多样性度量方法 ·1037·