In my former paper ＂A pre-order principle and set-valued Ekeland variational principle＂ （see [J. Math. Anal. Applo, 419, 904 937 （2014）]）, we established a general pre-order principle. From the pre-order princip...In my former paper ＂A pre-order principle and set-valued Ekeland variational principle＂ （see [J. Math. Anal. Applo, 419, 904 937 （2014）]）, we established a general pre-order principle. From the pre-order principle, we deduced most of the known set-valued Ekeland variational principles （denoted by EVPs） in set containing forms and their improvements. But the pre-order principle could not imply Khanh and Quy＇s EVP in [On generalized Ekeland＇s variational principle and equivalent formulations for set-valued mappings, J. Glob. Optim., 49, 381-396 （2011）], where the perturbation contains a weak T-function, a certain type of generalized distances. In this paper, we give a revised version of the pre-order principle. This revised version not only implies the original pre-order principle, but also can be applied to obtain the above Khanh and Quy＇s EVP. In particular, we give several new set-valued EVPs, where the perturbations contain convex subsets of the ordering cone and various types of generalized distances.展开更多
台区电力工单记录反映了台区运行工况和用户需求,是制定台区用电安全管理制度和满足台区用户用电需求的重要依据。针对台区电力工单高复杂性和强专业性给台区工单分类带来的难题,提出一种融合标签平滑(LS)与预训练语言模型的台区电力工...台区电力工单记录反映了台区运行工况和用户需求,是制定台区用电安全管理制度和满足台区用户用电需求的重要依据。针对台区电力工单高复杂性和强专业性给台区工单分类带来的难题,提出一种融合标签平滑(LS)与预训练语言模型的台区电力工单分类模型(MiniRBT-LSTM-GAT)。首先,利用预训练模型计算电力工单文本中的字符级特征向量表示;其次,采用双向长短期记忆网络(BiLSTM)捕捉电力文本序列中的依赖关系;再次,通过图注意力网络(GAT)聚焦对文本分类贡献大的特征信息;最后,利用LS改进损失函数以提高模型的分类精度。所提模型与当前主流的文本分类算法在农网台区电力工单数据集(RSPWO)、浙江省95598电力工单数据集(ZJPWO)和THUCNews(TsingHua University Chinese News)数据集上的实验结果表明,与电力审计文本多粒度预训练语言模型(EPAT-BERT)相比,所提模型在RSPWO、ZJPWO上的查准率和F1值分别提升了2.76、2.02个百分点和1.77、1.40个百分点;与胶囊神经网络模型BRsyn-caps(capsule network based on BERT and dependency syntax)相比,所提模型在THUCNews数据集上的查准率和准确率分别提升了0.76和0.71个百分点。可见,所提模型有效提升了台区电力工单分类的性能,并在THUCNews数据集上表现良好,验证了模型的通用性。展开更多
基金Supported by National Natural Science Foundation of China(Grant Nos.11471236 and 11561049)
文摘In my former paper ＂A pre-order principle and set-valued Ekeland variational principle＂ （see [J. Math. Anal. Applo, 419, 904 937 （2014）]）, we established a general pre-order principle. From the pre-order principle, we deduced most of the known set-valued Ekeland variational principles （denoted by EVPs） in set containing forms and their improvements. But the pre-order principle could not imply Khanh and Quy＇s EVP in [On generalized Ekeland＇s variational principle and equivalent formulations for set-valued mappings, J. Glob. Optim., 49, 381-396 （2011）], where the perturbation contains a weak T-function, a certain type of generalized distances. In this paper, we give a revised version of the pre-order principle. This revised version not only implies the original pre-order principle, but also can be applied to obtain the above Khanh and Quy＇s EVP. In particular, we give several new set-valued EVPs, where the perturbations contain convex subsets of the ordering cone and various types of generalized distances.
文摘台区电力工单记录反映了台区运行工况和用户需求,是制定台区用电安全管理制度和满足台区用户用电需求的重要依据。针对台区电力工单高复杂性和强专业性给台区工单分类带来的难题,提出一种融合标签平滑(LS)与预训练语言模型的台区电力工单分类模型(MiniRBT-LSTM-GAT)。首先,利用预训练模型计算电力工单文本中的字符级特征向量表示;其次,采用双向长短期记忆网络(BiLSTM)捕捉电力文本序列中的依赖关系;再次,通过图注意力网络(GAT)聚焦对文本分类贡献大的特征信息;最后,利用LS改进损失函数以提高模型的分类精度。所提模型与当前主流的文本分类算法在农网台区电力工单数据集(RSPWO)、浙江省95598电力工单数据集(ZJPWO)和THUCNews(TsingHua University Chinese News)数据集上的实验结果表明,与电力审计文本多粒度预训练语言模型(EPAT-BERT)相比,所提模型在RSPWO、ZJPWO上的查准率和F1值分别提升了2.76、2.02个百分点和1.77、1.40个百分点;与胶囊神经网络模型BRsyn-caps(capsule network based on BERT and dependency syntax)相比,所提模型在THUCNews数据集上的查准率和准确率分别提升了0.76和0.71个百分点。可见,所提模型有效提升了台区电力工单分类的性能,并在THUCNews数据集上表现良好,验证了模型的通用性。
