The transmission characteristics of the optical label switching system based on the FSK/ASK orthogonal modulation format is investigated. The factors that affect the transmission performance, such as the FSK tone spac...The transmission characteristics of the optical label switching system based on the FSK/ASK orthogonal modulation format is investigated. The factors that affect the transmission performance, such as the FSK tone space, dispersion compensation and coupler split ratio, are studied by numerical simulation. The proposed scheme is also experimentally demonstrated with a transmission of 155 Mbit/s FSK label combined with 10 Gbit/s ASK payload.展开更多
多示例多标记学习在多语义对象处理中克服了多示例学习和多标记学习的缺点,成功应用于文本分类、图像识别标注、基因数据分析等任务中.其中基于退化策略的多示例多标记学习算法,多利用K-Medoids聚类将多示例多标记退化成单示例多标记,...多示例多标记学习在多语义对象处理中克服了多示例学习和多标记学习的缺点,成功应用于文本分类、图像识别标注、基因数据分析等任务中.其中基于退化策略的多示例多标记学习算法,多利用K-Medoids聚类将多示例多标记退化成单示例多标记,但此种退化方式过于简化多语义和复杂语义的对象,并未考虑示例间的相关性,导致退化过程中的信息削弱甚至丢失.针对这一问题,提出了结合均值漂移的多示例多标记学习改进算法(MultiInstance Multi-Label with Mean Shift,MIMLMS),将高斯核函数和权值加入均值漂移中.权值的加入保证了示例之间的相关性得以保留,而将多示例集合加入高斯核函数就可利用核密度估计和梯度下降法求解退化过程最优解,最终以误差平方和为分类目标函数,建立多示例多标记分类模型.算法在基准的多示例多标记测试数据集中的实验结果,验证了算法的良好分类效果及算法的有效性和可靠性.展开更多
基金supported by the National Natural Science Foundation of China(Grant No 60677004)the National High Technology Research and Development Program of China(Grant No 2007AA01Z260)+4 种基金The project is also supported by the Key Project of Chinese Ministry of Education(Grant No 107011)the Key Laboratory of Broadband Optical Fiber Transmission and Communication Networks(UESTC)(Ministry of Education of China)Teaching and Scientific Research Foundation for the Returned Overseas Chinese Scholars(State Education Ministry of China)the Corporative Building Project of Beijing Educational Committee(Grant No XK100130737)the Program for New Century Excellent Talents in University of China(Grant No NECT-07-0111)
文摘The transmission characteristics of the optical label switching system based on the FSK/ASK orthogonal modulation format is investigated. The factors that affect the transmission performance, such as the FSK tone space, dispersion compensation and coupler split ratio, are studied by numerical simulation. The proposed scheme is also experimentally demonstrated with a transmission of 155 Mbit/s FSK label combined with 10 Gbit/s ASK payload.
文摘多示例多标记学习在多语义对象处理中克服了多示例学习和多标记学习的缺点,成功应用于文本分类、图像识别标注、基因数据分析等任务中.其中基于退化策略的多示例多标记学习算法,多利用K-Medoids聚类将多示例多标记退化成单示例多标记,但此种退化方式过于简化多语义和复杂语义的对象,并未考虑示例间的相关性,导致退化过程中的信息削弱甚至丢失.针对这一问题,提出了结合均值漂移的多示例多标记学习改进算法(MultiInstance Multi-Label with Mean Shift,MIMLMS),将高斯核函数和权值加入均值漂移中.权值的加入保证了示例之间的相关性得以保留,而将多示例集合加入高斯核函数就可利用核密度估计和梯度下降法求解退化过程最优解,最终以误差平方和为分类目标函数,建立多示例多标记分类模型.算法在基准的多示例多标记测试数据集中的实验结果,验证了算法的良好分类效果及算法的有效性和可靠性.
