ID 原文 译文
58478 在加利福尼亚大学欧文分校的机器学习库( UCI) 的 8 个公开数据集中进行对比实验结果表明基于最大相关系数的特征选择算法( NFCBF) 总体优于 FCBF 算法，它所选择出特征数比 FCBF 算法所选择特征数平均少了 3.625 个，分类准确率平均提高了 0.075% . Experimental results on machinelearning repository of university of california irvine( UCI) eight open datasets show that NFCBF algorithmoutperforms FCBF algorithm. The number of features selected by feature selection algorithm based onmaximum information coefficient ( NFCBF algorithm) is less than 3. 625 of the selected feature subset ofFCBF algorithm，and the classification accuracy is improved by 0. 075% .
58479 与互信息最大算法( MIM) 、最少的绝对收缩和选择算法( Lasso) 和岭算法( Ｒidge) 等相比也具有明显的优势. NFCBF algorithm gives betterperformance than mutual information maximization algorithm( MIM) ，Least absolute shrinkage and selec?tion operator algorithm( Lasso) and Ｒidge algorithm.
58480 为了解决当前车联网中节点处理任务的时效性问题，基于最优节点选取及任务卸载理论，提出了一种新的任务卸载决策方案. In order to solve the problem of the timeliness of nodes processing tasks in the internet of vehi?cles，a new task offloading decision-making scheme is proposed based on the optimal node selection andtask offloading theory.
58481 该方案对车联网场景下的任务卸载进行建模，构造出具有指向性的节点分布模型，利用最优节点选取算法对该模型进行节点预选取，通过一次或二次任务卸载预判机制，实现节点对任务的卸载决策. First of all，the task offloading model is built and a directional node distributionmodel is constructed. Furthermore，the node is pre-selected by using the optimal node selection algo?rithm. Finally，the task offloading decision is achieved by once or second task offloading pre-judgmentmechanism.
58482 与传统的任务卸载决策相比，所提方案在任务卸载全过程中耗时更少，实时性更高. Compared with the traditional task offloading decision-making，the proposed solution takesless time，and the task Ｒeal-time is more higher.
58483 仿真结果验证了所提方法的有效性. Simulation results verify the effectiveness of the proposed method.
58484 简要介绍了设计模式自动识别领域的发展历程;从基于图论、形式化技术、软件度量和人工智能的设计模式识别四大类方法出发，对该领域的研究进行了综述; The development history of the field of automatic design pattern detection was briefly intro?duced; the research of this field was respectively reviewed from four major categories of approaches:graph theory-based，formal technologies-based，software metrics-based and artificial intelligence-baseddesign pattern detection; the four major categories of approaches are compared through 3 indexes:
58485 通过适用的模式类型、时间性能和能否完全自动化 3 个指标对四大类方法进行了比较; appli?cable pattern type ( s) ，time performance and possibility of fully automated; finally the problems andshortcomings of the current research were analyzed，and the future research directions were pointed out.
58486 最后分析了当前研究存在的问题与不足，并指出了今后的研究方向.对各方法适用的模式类型、优缺点及内在联系进行了分析和讨论，从宏观上阐述了该领域的研究进展. This paper analyzes and discusses the applicable pattern type( s) ，the advantages / disadvantages and theinner connections of the various approaches，and expounds the research progress of this field from themacro level.
58487 针对现有无线医疗传感器网络环境下身份认证协议在登录和密码更新阶段效率低下的问题，提出了一种基于椭圆曲线加密的身份认证协议. Due to the inefficiency of the existing identity authentication protocol during the login andpassword update phase in the existing wireless medical sensor network environment，an elliptic curvecryptosystem based authentication protocol suing smart card was presented.