A new incremental support vector machine （SVM） algorithm is proposed which is based on multiple kernel learning. Through introducing multiple kernel learning into the SVM incremental learning, large scale data set l...A new incremental support vector machine （SVM） algorithm is proposed which is based on multiple kernel learning. Through introducing multiple kernel learning into the SVM incremental learning, large scale data set learning problem can be solved effectively. Furthermore, different punishments are adopted in allusion to the training subset and the acquired support vectors, which may help to improve the performance of SVM. Simulation results indicate that the proposed algorithm can not only solve the model selection problem in SVM incremental learning, but also improve the classification or prediction precision.展开更多
The scheme for tracking maneuvering target based on neural fuzzy network with incremental neural learning is proposed. When tracked target maneuver occurs, the scheme can detect maneuver immediately and estimate the m...The scheme for tracking maneuvering target based on neural fuzzy network with incremental neural learning is proposed. When tracked target maneuver occurs, the scheme can detect maneuver immediately and estimate the maneuver value accurately , then the tracking filter can be compensated correctly and duly by the estimated maneuver value. When environment changes, neural fuzzy network with incremental neural learning （INL-SONFIN） can find its optimal structure and parameters automatically to adopt to changed environment. So, it always produce estimated output very close to the true maneuver value that leads to good tracking performance and avoids misstracking. Simulation results show that the performance is superior to the traditional schemes and the scheme can fit changed dynamic environment to track maneuvering target accurately and duly.展开更多
A new incremental clustering framework is presented, the basis of which is the induction as inverted deduction. Induction is inherently risky because it is not truth-preserving. If the clustering is considered as an i...A new incremental clustering framework is presented, the basis of which is the induction as inverted deduction. Induction is inherently risky because it is not truth-preserving. If the clustering is considered as an induction process, the key to build a valid clustering is to minimize the risk of clustering. From the viewpoint of modal logic, the clustering can be described as Kripke frames and Kripke models which are reflexive and symmetric. Based on the theory of modal logic, its properties can be described by system B in syntax. Thus, the risk of clustering can be calculated by the deduction relation of system B and proximity induction theorem described. Since the new proposed framework imposes no additional restrictive conditions of clustering algorithm, it is therefore a universal framework. An incremental clustering algorithm can be easily constructed by this framework from any given nonincremental clustering algorithm. The experiments show that the lower the a priori risk is, the more effective this framework is. It can be demonstrated that this framework is generally valid.展开更多
社交网络中,节点间存在多种关系类型,节点数量会随着时间的推移而变化,这种异质性和动态性给链路预测任务带来极大的挑战。因此,本文提出一种基于增量学习的社交网络链路预测方法(incremental learning social networks link prediction...社交网络中,节点间存在多种关系类型,节点数量会随着时间的推移而变化,这种异质性和动态性给链路预测任务带来极大的挑战。因此,本文提出一种基于增量学习的社交网络链路预测方法(incremental learning social networks link prediction,IL-SNLP)。通过对网络进行分层,使每一层网络只包含一种关系类型,以更好地获取节点在每种关系类型下的语义信息;针对网络的动态性,利用时序随机游走捕获社交网络中的局部结构信息和时序信息;针对增量数据,采用增量式更新随机游走策略对历史随机游走序列进行更新。通过增量式skip-gram模型从随机游走序列中提取新出现节点的特征,并进一步更新历史节点的特征;针对网络的异质性,采用概率模型提取不同关系类型之间的因果关系关联程度,并将其作用于每一层的节点特征,以改善不同关系层下节点特征表现能力;利用多层感知机构建节点相互感知器,挖掘节点间建立连接时的相互贡献,实现更高的链路预测准确率。实验结果表明,在3个真实的社交网络数据集上,IL-SNLP方法的ROC曲线下的面积(AUC)和F1分数比基线方法分别提高了10.08%~67.60%和1.76%~64.67%,提升了预测性能;对于增量数据,只需要少次迭代就能保持预测模型的性能,提高了模型训练的速度;与未采用增量学习技术的IL-SNLP−方法相比,IL-SNLP方法在时间效率上提升了30.78%~257.58%,显著缩短了模型的运行时长。展开更多
从遥感影像上自动解译铁路设计控制要素是实现“一键成图”的关键,但深度学习遥感影像智能解译需要大量标注样本。依据铁路线路设计原则,提出一种多源遥感数据的设计控制要素智能解译样本库构建方法。首先,基于数字正射影像图(Digital O...从遥感影像上自动解译铁路设计控制要素是实现“一键成图”的关键,但深度学习遥感影像智能解译需要大量标注样本。依据铁路线路设计原则,提出一种多源遥感数据的设计控制要素智能解译样本库构建方法。首先,基于数字正射影像图(Digital Orthophoto Map,DOM)、数字线划地图(Digital Line Graphic,DLG)和激光雷达(Light Detection and Ranging,Lidar)点云多源数据自动生成初始样本;其次,基于增量主动学习迭代方法对初始样本进行优化,达到高质量、全面覆盖铁路沿线的目的;然后,以长赣铁路为例,构建以铁路沿线周边房屋、道路、水体和植被四类铁路线路设计控制要素为重点的高分辨率智能解译样本数据库——铁路线路设计控制要素智能解译样本库(Wuhan University Sample Database of Control Elements of Railway Route Design,WHU-RRDSD),其地面分辨率为0.1 m,样本总数超过20万张;最后,为验证样本库的可用性,分别从定性评价、定量评价以及其他场景应用案例三方面进行详细验证,结果表明,基于房屋、道路、水体和植被四类样本库的IoU评价指标分别为84.43%、82.38%、90.19%、90.28%,表现出优异的解译效果;基于WHU-RRDSD训练得到的智能模型迁移至宜涪高铁场景中房屋、道路、水体和植被要素的解译,验证样本库在其他场景的可用性;简要介绍基于WHU-RRDSD样本库进行的高分辨率遥感图像弱监督建筑提取和高分辨率遥感图像地物分类两个应用案例,进一步验证本文方法所构建样本库可用性。展开更多
基金supported by the National Natural Science Key Foundation of China(69974021)
文摘A new incremental support vector machine （SVM） algorithm is proposed which is based on multiple kernel learning. Through introducing multiple kernel learning into the SVM incremental learning, large scale data set learning problem can be solved effectively. Furthermore, different punishments are adopted in allusion to the training subset and the acquired support vectors, which may help to improve the performance of SVM. Simulation results indicate that the proposed algorithm can not only solve the model selection problem in SVM incremental learning, but also improve the classification or prediction precision.
基金This project was supported by Spaceflight Support Fund ( HIT01) and the Spaceflight Science Project Group
文摘The scheme for tracking maneuvering target based on neural fuzzy network with incremental neural learning is proposed. When tracked target maneuver occurs, the scheme can detect maneuver immediately and estimate the maneuver value accurately , then the tracking filter can be compensated correctly and duly by the estimated maneuver value. When environment changes, neural fuzzy network with incremental neural learning （INL-SONFIN） can find its optimal structure and parameters automatically to adopt to changed environment. So, it always produce estimated output very close to the true maneuver value that leads to good tracking performance and avoids misstracking. Simulation results show that the performance is superior to the traditional schemes and the scheme can fit changed dynamic environment to track maneuvering target accurately and duly.
基金supported by the National High-Tech Research and Development Program of China(2006AA12A106).
文摘A new incremental clustering framework is presented, the basis of which is the induction as inverted deduction. Induction is inherently risky because it is not truth-preserving. If the clustering is considered as an induction process, the key to build a valid clustering is to minimize the risk of clustering. From the viewpoint of modal logic, the clustering can be described as Kripke frames and Kripke models which are reflexive and symmetric. Based on the theory of modal logic, its properties can be described by system B in syntax. Thus, the risk of clustering can be calculated by the deduction relation of system B and proximity induction theorem described. Since the new proposed framework imposes no additional restrictive conditions of clustering algorithm, it is therefore a universal framework. An incremental clustering algorithm can be easily constructed by this framework from any given nonincremental clustering algorithm. The experiments show that the lower the a priori risk is, the more effective this framework is. It can be demonstrated that this framework is generally valid.
文摘社交网络中,节点间存在多种关系类型,节点数量会随着时间的推移而变化,这种异质性和动态性给链路预测任务带来极大的挑战。因此,本文提出一种基于增量学习的社交网络链路预测方法(incremental learning social networks link prediction,IL-SNLP)。通过对网络进行分层,使每一层网络只包含一种关系类型,以更好地获取节点在每种关系类型下的语义信息;针对网络的动态性,利用时序随机游走捕获社交网络中的局部结构信息和时序信息;针对增量数据,采用增量式更新随机游走策略对历史随机游走序列进行更新。通过增量式skip-gram模型从随机游走序列中提取新出现节点的特征,并进一步更新历史节点的特征;针对网络的异质性,采用概率模型提取不同关系类型之间的因果关系关联程度,并将其作用于每一层的节点特征,以改善不同关系层下节点特征表现能力;利用多层感知机构建节点相互感知器,挖掘节点间建立连接时的相互贡献,实现更高的链路预测准确率。实验结果表明,在3个真实的社交网络数据集上,IL-SNLP方法的ROC曲线下的面积(AUC)和F1分数比基线方法分别提高了10.08%~67.60%和1.76%~64.67%,提升了预测性能;对于增量数据,只需要少次迭代就能保持预测模型的性能,提高了模型训练的速度;与未采用增量学习技术的IL-SNLP−方法相比,IL-SNLP方法在时间效率上提升了30.78%~257.58%,显著缩短了模型的运行时长。
文摘从遥感影像上自动解译铁路设计控制要素是实现“一键成图”的关键,但深度学习遥感影像智能解译需要大量标注样本。依据铁路线路设计原则,提出一种多源遥感数据的设计控制要素智能解译样本库构建方法。首先,基于数字正射影像图(Digital Orthophoto Map,DOM)、数字线划地图(Digital Line Graphic,DLG)和激光雷达(Light Detection and Ranging,Lidar)点云多源数据自动生成初始样本;其次,基于增量主动学习迭代方法对初始样本进行优化,达到高质量、全面覆盖铁路沿线的目的;然后,以长赣铁路为例,构建以铁路沿线周边房屋、道路、水体和植被四类铁路线路设计控制要素为重点的高分辨率智能解译样本数据库——铁路线路设计控制要素智能解译样本库(Wuhan University Sample Database of Control Elements of Railway Route Design,WHU-RRDSD),其地面分辨率为0.1 m,样本总数超过20万张;最后,为验证样本库的可用性,分别从定性评价、定量评价以及其他场景应用案例三方面进行详细验证,结果表明,基于房屋、道路、水体和植被四类样本库的IoU评价指标分别为84.43%、82.38%、90.19%、90.28%,表现出优异的解译效果;基于WHU-RRDSD训练得到的智能模型迁移至宜涪高铁场景中房屋、道路、水体和植被要素的解译,验证样本库在其他场景的可用性;简要介绍基于WHU-RRDSD样本库进行的高分辨率遥感图像弱监督建筑提取和高分辨率遥感图像地物分类两个应用案例,进一步验证本文方法所构建样本库可用性。
