This paper addresses the observer design problem for a class of nonlinear descriptor systems whose nonlinear terms are slope-restricted. The full-order observer is formulated as a nonlinear descriptor system. A linear...This paper addresses the observer design problem for a class of nonlinear descriptor systems whose nonlinear terms are slope-restricted. The full-order observer is formulated as a nonlinear descriptor system. A linear matrix inequality （LMI） condition is derived to construct the full-order observer. The existence and uniqueness of the solution to the obtained observer system are guaranteed. Furthermore, under the same LMI condition and a common assumption, a reduced-order observer is designed. Finally, the design methods are reduced to a strict LMI problem and illustrated by a numerical example.展开更多
针对推荐系统的数据稀疏性导致的推荐效果不佳的问题,提出一种基于评分填充与信任信息的混合推荐的算法RTWSO(Real-value user item restricted Boltzmann machine Trust WSO)。首先,使用改进的受限玻尔兹曼机模型对评分矩阵进行填充,...针对推荐系统的数据稀疏性导致的推荐效果不佳的问题,提出一种基于评分填充与信任信息的混合推荐的算法RTWSO(Real-value user item restricted Boltzmann machine Trust WSO)。首先,使用改进的受限玻尔兹曼机模型对评分矩阵进行填充,以缓解评分矩阵的稀疏性问题;其次,从信任关系中提取信任与被信任关系,并通过基于矩阵分解的隐含信任关系相似度来解决信任信息稀疏的问题,而且对原有算法进行了包含信任信息的修正,以提高推荐准确度;最后,通过加权Slope One(WSO)算法对矩阵填充与信任相似度信息加以整合,并对评分数据进行预测。在Epinions与Ciao数据集中验证算法性能,可见所提出混合推荐算法较组成算法在推荐准确度上提升3%以上,较现有社会化推荐算法SocialIT(Social recommendation algorithm based on Implict similarity in Trust)在推荐准确度上提升1.2%以上。实验结果表明,所提出的基于评分填充与信任信息的混合推荐算法在一定程度上提高了推荐准确度。展开更多
基金supported by National Basic Research Program of China (973 Program) (No. 2009CB320601)National Natural Science Foundation of China (No. 60904009)Fundamental Research Funds for the Central Universities (No. 100406010, No. 090408001)
文摘This paper addresses the observer design problem for a class of nonlinear descriptor systems whose nonlinear terms are slope-restricted. The full-order observer is formulated as a nonlinear descriptor system. A linear matrix inequality （LMI） condition is derived to construct the full-order observer. The existence and uniqueness of the solution to the obtained observer system are guaranteed. Furthermore, under the same LMI condition and a common assumption, a reduced-order observer is designed. Finally, the design methods are reduced to a strict LMI problem and illustrated by a numerical example.
文摘针对推荐系统的数据稀疏性导致的推荐效果不佳的问题,提出一种基于评分填充与信任信息的混合推荐的算法RTWSO(Real-value user item restricted Boltzmann machine Trust WSO)。首先,使用改进的受限玻尔兹曼机模型对评分矩阵进行填充,以缓解评分矩阵的稀疏性问题;其次,从信任关系中提取信任与被信任关系,并通过基于矩阵分解的隐含信任关系相似度来解决信任信息稀疏的问题,而且对原有算法进行了包含信任信息的修正,以提高推荐准确度;最后,通过加权Slope One(WSO)算法对矩阵填充与信任相似度信息加以整合,并对评分数据进行预测。在Epinions与Ciao数据集中验证算法性能,可见所提出混合推荐算法较组成算法在推荐准确度上提升3%以上,较现有社会化推荐算法SocialIT(Social recommendation algorithm based on Implict similarity in Trust)在推荐准确度上提升1.2%以上。实验结果表明,所提出的基于评分填充与信任信息的混合推荐算法在一定程度上提高了推荐准确度。
