检索结果-南通市图书馆

Model-independent Veltman condition,naturalness and the little hierarchy problem

在线全文

学校读者我要写书评

暂无评论

Chinese Physics C 2022年第1期46卷 31-42页

作者： Fayez Abu-Ajamieh LUPM UMR5299 Universitéde Montpellier34095 MontpellierFrance

We adopt a bottom-up Effective Field Theory(EFT)approach to derive a model-independent Veltman condition to cancel out the quadratic divergences in the Higgs *** show using the equivalence theorem that all the deviations in the Higgs couplings to the W and Z from the SM predictions should *** argue based on tree-level unitarity that any new physics that naturally cancels out the quadratic divergences should be ≤19 *** show that the level of fine-tuning required is unless the O(0:1%-1%) UV sector has a symmetry that forces the satisfaction of the model-independent Veltman condition,in which case all fine-tuning is *** also conjecture that,if no new physics that couples to the Higgs is observed up to~19 TeV,or if the Higgs couplings to the SM particles conform to the SM predictions,then the Higgs either does not couple to any UV sector or is fine-tuned.

关键词： effective field theory Higgs little hierarchy problem quadratic divergence fine-tuning unitarity byeond the standard model

Nonlinear system identification using deep belief network based on PLSR 36

在线全文

学校读者我要写书评

暂无评论

Nonlinear system identification using deep belief network ba...

BERT-TECNN模型的文本分类方法研究

第36届中国控制会议

作者： Wang Gongming Li Wenjing Qiao Junfei Wu Guandi Faculty of Information Technology Beijing University of Technology Beijing Key Laboratory of Computational Intelligence and Intelligent System Technical Test Center of Shengli Oilfield Branch Sinopec

ISBN: (纸本)9781538629185

Deep learning has been successfully applied into pattern recognition due to its deep architecture and effective unsupervised learning,and deep belief network（DBN） is a popular model based on deep learning *** this paper,a DBN identification model based on partial least square regression（PLSR）,named PLSR-DBN,is proposed for nonlinear system *** order to improve the identification accuracy,PLSR is introduced into the supervised fine-tuning of DBN to elimate the overfitting and local minimum resulted from gradients-based learning,and contrastive divergence（CD） algorithm is used in unsupervised ***,the proposed PLSR-DBN is tested on a benchmark nonlinear *** experiment results show that the proposed PLSR-DBN has a better performance on nonlinear system identification than other similar methods.

关键词： Deep learning Deep belief network Nonliear system identification fine-tuning Partial least square regression

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

计算机工程与应用 2021年第18期57卷 186-193页

作者：李铁飞生龙吴迪河北工程大学信息与电气工程学院河北邯郸056107 河北工程大学河北省安防信息感知与处理重点实验室河北邯郸056107

由于Bert-base,Chinese预训练模型参数巨大,在做分类任务微调时内部参数变化较小,易产生过拟合现象,泛化能力弱,且该模型是以字为单位进行的预训练,包含词信息量较少。针对这些问题,提出了BERT-TECNN模型,模型使用Bert-base,Chinese模... 详细信息

由于Bert-base,Chinese预训练模型参数巨大,在做分类任务微调时内部参数变化较小,易产生过拟合现象,泛化能力弱,且该模型是以字为单位进行的预训练,包含词信息量较少。针对这些问题,提出了BERT-TECNN模型,模型使用Bert-base,Chinese模型作为动态字向量模型,输出包含深度特征信息的字向量,Transformerencoder层再次对数据进行多头自注意力计算,提取特征信息,以提高模型的泛化能力,CNN层利用不同大小卷积核,捕捉每条数据中不同长度词的信息,最后应用softmax进行分类。该模型与Word2Vec+CNN、Word2Vec+BiLSTM、Elmo+CNN、BERT+CNN、BERT+BiLSTM、BERT+Transformer等深度学习文本分类模型在三种数据集上进行对比实验,得到的准确率、精确率、召回率、F1测度值均为最高。实验表明该模型有效地提取了文本中字词的特征信息,优化了过拟合问题,提高了泛化能力。

关键词： bert transformer encoder CNN 文本分类 fine-tuning self-attention 过拟合

Magnetic Interactions Switching in Two Dinuclear Dy（Ⅲ） Single-Molecule Magnets via Subtle Structural Distortion

在线全文

学校读者我要写书评

暂无评论

Magnetic Interactions Switching in Two Dinuclear Dy（Ⅲ） Si...

中国化学会第九届全国无机化学学术会议

作者： Kun Zhang Yao-Yu Wang Key Laboratory of Synthetic and Natural Functional Molecule Chemistry of the Ministry of Education Shaanxi Key Laboratory of Physico-Inorganic ChemistryCollege of Chemistry ＆ Materials ScienceNorthwest University

In the field of coordination and supramolecular chemistry,lanthanide complexes have attracted intense interest of scientists over recent years for their interesting optical and magnetic *** particular,numerous experimental and theoretical studies on lanthanide coordination complexes have been explored since Ishikawa et *** a terbium phthalocyanine complex which could exhibit single-molecule magnet（SMM）*** discrete molecules can retain spin information over long periods of time and behave like nanoscopic classical magnets at low temperature,suggesting possible applications in molecular spintronics,ultra-high density information storage and quantum information processing.1The study for structure-property relationship is the eternal theme of modern *** whose goals are the design of new matters exhibiting desired physical or chemical properties,rather than only being satisfied with creating new *** recent years,the detailed analysis of the correlation between structure and magnetic properties has been established for transition metal clusters,by contrast,the study of magneto-structural correlation for lanthanide complexes is a close blank *** is mainly due to the fact that,ferromagnetically coupled polynuclear lanthanide complexes are extremely rare,and no suitable models are supported.2In this work,we have structurally and magnetically characterized two dinuclear dysprosium complexes,1and 2（Fig.1）.The two complexes display only small structural variations for the geometry of{Dy2O2}cores but remarkably different magnetic *** slight variations for the ligands but same coordination atoms lead a change from antiferromagnetic to ferromagnetic interactions in those two *** current susceptibilities reveal that both of them exhibit single-molecule magnet *** two centrosymmetric{Dy2}complexes support ideal models to elucidate the magneto-structural correlation for lanthanide comple

关键词： Dysprosium fine-tuning Magneto-Structural Correlation Single-Molecule Magnets

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

仓储环境下基于深度学习的物体识别方法研究

北京信息科技大学学报（自然科学版） 2018年第1期33卷 60-65页

作者：金秋李天剑北京信息科技大学机电工程学院北京100192

针对仓储环境下叉车机器人物体识别的应用场景,提出一种基于Faster-RCNN优化和改进后的物体识别算法。通过对Faster-RCNN模型进行微调(fine-tuning),完成对托盘、货物、人以及叉车等物体的识别,同时优化了训练过程,使得网络最后达到最... 详细信息

针对仓储环境下叉车机器人物体识别的应用场景,提出一种基于Faster-RCNN优化和改进后的物体识别算法。通过对Faster-RCNN模型进行微调(fine-tuning),完成对托盘、货物、人以及叉车等物体的识别,同时优化了训练过程,使得网络最后达到最优。同时通过对不同共享卷积层模型下的Faster-RCNN进行比较,最后得到最优的Faster-RCNN模型为ZF+RPN模型。实验表明,改进和优化后的算法对仓储环境下的物体检测的准确率达到90%,测试的帧率为33.3fps,基本满足叉车机器人对物体检测实时性和准确性的要求。

关键词：深度学习 Faster-RCNN fine-tuning RPN模型共享卷积层

Medical care service fee makes the value of doctor＇s work recognized

在线全文

学校读者我要写书评

暂无评论

Chinese Medical Journal 2014年第13期127卷 2556-2556页

作者： Zhou Haijiang Lan Tianfei Yan Yong Li Chunsheng Capital Med Univ Beijing Chao Yang Hosp Dept Emergency Med Beijing 100020 Peoples R China Chengde Med Coll Affiliated Hosp Dept Dermatol Chengde 067000 Hebei Peoples R China Capital Med Univ Beijing Shijitan Hosp Dept Urol Beijing 100038 Peoples R China

To the editor： Beijing has begun implementing its new public hospital reform programs since July 1, 2012.1 According to the new policies, patients are charged a medical care service fee which varies according to doctor＇s seniority while the 15 percent of drug markup is eliminated, aiming at reducing hospitals＇ dependency on profits from drug prescriptions and fine-tuning the structure of hospitals＇ incomes.

关键词： doctor's work recognized fine-tuning