针对现有基于文本的命名实体识别方法难以有效利用视觉信息,且主流多模态命名实体识别(multimodal named entity recognition,MNER)方法存在跨模态语义关联挖掘不足、异构数据融合能力有限、易受模态语义鸿沟影响等问题,提出一种基于显...针对现有基于文本的命名实体识别方法难以有效利用视觉信息,且主流多模态命名实体识别(multimodal named entity recognition,MNER)方法存在跨模态语义关联挖掘不足、异构数据融合能力有限、易受模态语义鸿沟影响等问题,提出一种基于显隐式双路径融合的多模态命名实体识别模型DPF-MNER(dual-path fusion MNER)。该模型引入双路径融合机制实现跨模态深度对齐:在显式路径中,构建目标实体-词汇关系图,明确建模文本实体与图像区域间的语义对应关系;在隐式路径中,设计基于动量对比学习的难样本对齐机制,通过动量更新维护跨模态记忆库,引导模型在共享语义空间中拉近相关图文对、推远不相关图文对,缓解模态偏差。在构建的军事领域专用数据集ME-MNER与公开数据集Twitter-2017上的实验结果表明,DPF-MNER在F1指标上分别达到87.05%和86.35%,验证了该方法在提升实体识别精度与模型泛化能力方面的有效性。展开更多
In this paper we study the algorithms and their parallel implementation for solving large-scale generalized eigenvalue problems in modal analysis.Three predominant subspace algorithms,i.e.,Krylov-Schur method,implicit...In this paper we study the algorithms and their parallel implementation for solving large-scale generalized eigenvalue problems in modal analysis.Three predominant subspace algorithms,i.e.,Krylov-Schur method,implicitly restarted Arnoldi method and Jacobi-Davidson method,are modified with some complementary techniques to make them suitable for modal analysis.Detailed descriptions of the three algorithms are given.Based on these algorithms,a parallel solution procedure is established via the PANDA framework and its associated eigensolvers.Using the solution procedure on a machine equipped with up to 4800processors,the parallel performance of the three predominant methods is evaluated via numerical experiments with typical engineering structures,where the maximum testing scale attains twenty million degrees of freedom.The speedup curves for different cases are obtained and compared.The results show that the three methods are good for modal analysis in the scale of ten million degrees of freedom with a favorable parallel scalability.展开更多
基金supported by grants from the Spanish Government Grants BSO2000-0108-C02-01SEJ2004-00752/PSIC+1 种基金Integrated Action HF2006-0037 and by grants from the Comunidad de Madrid (Ref.:06/HSE/2004 and MULTIMAG: 2006/ BIO-170)JM were supported by a predoctoral grant from Ministerio de Educación y Ciencia (Ref.:AP2003-0639)
基金supported by the National Defence Basic Fundamental Research Program of China(Grant No.C1520110002)the Fundamental Development Foundation of China Academy Engineering Physics(Grant No.2012A0202008)
文摘In this paper we study the algorithms and their parallel implementation for solving large-scale generalized eigenvalue problems in modal analysis.Three predominant subspace algorithms,i.e.,Krylov-Schur method,implicitly restarted Arnoldi method and Jacobi-Davidson method,are modified with some complementary techniques to make them suitable for modal analysis.Detailed descriptions of the three algorithms are given.Based on these algorithms,a parallel solution procedure is established via the PANDA framework and its associated eigensolvers.Using the solution procedure on a machine equipped with up to 4800processors,the parallel performance of the three predominant methods is evaluated via numerical experiments with typical engineering structures,where the maximum testing scale attains twenty million degrees of freedom.The speedup curves for different cases are obtained and compared.The results show that the three methods are good for modal analysis in the scale of ten million degrees of freedom with a favorable parallel scalability.
