基于ROC的三元再编码研究

雷蕾; 王晓丹; 罗玺

doi:10.11999/JEIT151343

基于ROC的三元再编码研究

doi: 10.11999/JEIT151343

1.
(空军工程大学防空反导学院西安 710051) ②(空军工程大学信息与导航学院西安 710077)

基金项目:

国家自然科学基金(61273275, 61503407)

计量
- 文章访问数: 1152
- HTML全文浏览量: 88
- PDF下载量: 340
- 被引次数: 0
出版历程
- 收稿日期: 2015-12-01
- 修回日期: 2016-06-06
- 刊出日期: 2016-10-19

Recoding Error-correcting Output Codes Based on Receiver Operating Characteristics

1.
(Institute of Air and Missile Defense, Aire Force Engineering University, Xi&rsquo
2.
(Institute of Information and Navigation, Aire Force Engineering University, Xi&rsquo

Funds:

The National Natural Science Foundation of China (61273275, 61503407)

摘要

摘要: 针对三元编码矩阵中基分类器不包含被忽略样本类别先验知识的问题，该文提出一种基于接收机工作特性(ROC)曲线的矩阵再编码方法。首先基于ROC曲线寻找构造拒绝域的阈值对，从而获得最优分类器；然后利用最优分类器对训练样本中被忽略的类别进行分类，将经典的二值输出变为三值输出，从而对初始编码矩阵的码元0进行重新编码。在解码阶段，采用经典的汉明距离解码方法对未知样本进行决策。该方法能够避免基分类器的二次训练，适用于任意的三元纠错输出编码，具有良好的普适性和实用性。基于人工和UCI公共数据集的实验结果表明该方法简单高效，在不增加训练时间的基础上，能够提高解码的速度和精度，促进分类效果的提升。
- 三元纠错输出编码 /
- 二次编码 /
- 最优分类器 /
- 拒绝域 /
- 接收机工作特性
Abstract: As to the problem that the base classifiers in ternary Error Correcting Output Codes (ECOC) matrix do not contain the prior information of classes which are ignored in binary splits, a new recoding ECOC based on Receiver Operating Characteristic (ROC) curve is presented. To recode the ternary matrix, the two thresholds of reject region are obtained based on ROC to build the optimal classifiers. Then, the optimal classifiers are used to classify the ignored classes based on bipartition in training phase. In so doing, the classical two-symbol output expands to three-symbol to recode the zeros. Finally, the Hamming decoding strategy is adopted for decision in decoding. This method can avoid a second training and is applied to any kind of ternary matrix. The experiments based on Synthetic and UCI datasets validate the better efficiency and remarkable promotion without increasing training complexity of the proposed approach.
- Ternary error-correcting output codes /
- Recoding /
- Optimal classifier /
- Reject option /
- Receiver Operating Characteristics (ROC)

HTML全文

参考文献(20)

DIETTERICH T G and BAKIRI G. Solving multi-class learning problems via error-correcting output codes[J]. Journal of Artificial Intelligence Research, 1995, 34(2): 263-286. doi: 10.1613/jair.105.

PHYO K S, JIAN G W, and EAM K T. Facial age range estimation with extreme learning machines[J]. Neurocomputing, 2015, 149A: 364-372. doi: 10.1016/ j.neucom.2014.03.074.

ELIF D. ECG beats classification using multiclass support vector machines with error correcting output codes[J]. Digital Signal Processing, 2007, 45(17): 675-684. doi: 10.1016/j.dsp. 2006.11.009.

SERGIO E, DAVID M, ELOI P, et al. Online error correcting output codes[J]. Pattern Recognition Letters, 2011, 32(3): 458-467. doi: 10.1016/j.patrec.2010.11.005.

ERIN L A, ROBERT E S, YORAM S, et al. Reducing multiclass to binary: a unifying approach for margin classifiers[J]. Journal of Machine Learning Research, 2000, 39(1): 113-141. doi: 10.1162/15324430152733133.

SERGIO E, ORIOL P, and PETIA R. Separability of ternary error-correcting output codes[J]. Pattern Recognition Letters, 2009, 30(5): 285-297. doi: 10.1016/j.patrec.2008.10.002.

SERGIO E, DAVID M J T, ORIOL P, et al. Subclass problem-dependent design for error-correcting output codes [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, 30(6): 1041-1054. doi: 10.1109/TPAMI. 2008.38.

周进登, 王晓丹, 周红健. 基于混淆矩阵的自适应纠错输出编码多类分类方法[J].系统工程与电子技术, 2012, 34(7): 220-226. doi: 10.3969/j.issn.1001-506X.2012.07.38.

ZHOU Jindeng, WANG Xiaodan, and ZHOU Hongjian. Multiclass classification of adaptive error-correcting output codes based on confusion matrix[J]. Systems Engineering and Electronics, 2012, 34(7): 220-226. doi: 10.3969/j.issn. 1001-506X.2012.07.38.

WANG Y, CHEN S C, and XUE H. Can under-exploited structure of original-classes help ECOC-based multi-class classification?[J]. Neurocomputing, 2012, 89: 158-167. doi: 10.1016/j.neucom.2012.02.035.

SERGIO E, ORIOL P, and PETIA R. Re-coding ECOCs without Re-training[J]. Pattern Recognition Letters, 2010, 31(7): 555-562. doi: 10.1016/j.patrec.2009.12.002.

MIGUEL A B, SERGIO E, XAVIER B, et al. On the design of an ECOC-compliant genetic algorithm[J]. Pattern Recognition, 2014, 47(2): 865-884. doi: 10.1016/j.patcog. 2013.06.019.

FRANCESCO C, ORIOL P, and PETIA R. ECOC-DRF: discriminative random fields based on error correcting output codes[J]. Pattern Recognition, 2014, 47(6): 2193-2204. doi: 10.1016/j.patcog.2013.12.007.

MIKEL G, ALBERTO F, EDURNE B, et al. DRCW-OVO: distance-based relative competence weighting combination for one-vs-one strategy in multi-class problems[J]. Pattern Recognition, 2015, 48(1): 28-52. doi: 10.1016/j.patcog. 2014.07.023.

LEI L, WANG X D, LUO X, et al. Hierarchical error-correcting output codes based on SVDD[J]. Pattern Analysis and Applications, 19(1): 163-171. doi: 10.1007/ s10044-015-0455-5.

TADEUSZ P. On the use of ROC analysis for the optimization of abstaining classifiers[J]. Machine Learning, 2007, 68(2): 137-169. doi: 10.1007/s10994-007-5013-y.

ZHOU J D, and WANG X D. Research on the unbiased probability estimation of error-correcting output coding[J]. Pattern Recognition, 2011, 44(7): 1552-1565. doi: 10.1016/ j.patcog.2010.12.020.

ZHOU J D, YUN Y，ZHANG J M, et al. Constructing ECOC based on confusion matrix for multiclass learning problems[J]. Science China Information Sciences, 2016, 59(1): 1-14. doi: 10.10071/s11432-015-5321-y.

邹洪侠, 秦峰. 二类分类器的ROC曲线生成算法[J]. 计算机技术与发展, 2009, 19(6): 109-112.

ZOU Hongxia and QIN Feng, Algorithm for generating ROC curve of two-classifier[J]. Computer Technology and Development, 2009, 19(6): 109-112.

施引文献

资源附件(0)

访问统计