In this paper we present a comparative analysis of global frequency and local deformation data for a large concrete bridge. The asymptotic probability distributions of the central statistics are presented, and compare...In this paper we present a comparative analysis of global frequency and local deformation data for a large concrete bridge. The asymptotic probability distributions of the central statistics are presented, and compared with empirical bootstrap estimates. Bootstrapped distributions are calculated from reference data obtained during 1999–2000 and used to develop change-point alarm criteria for the structure, using reasonable sensitivity measures developed from FEM simulations and structural analysis. The implications of the frequency data are discussed in conjunction with the strain and displacement measurements in order to discern if the load carrying capacity of the bridge has been affected. The critical need for more advanced temperature compensation models for large structures continually in thermal disequilibrium is discussed.展开更多
针对光伏出力和电动汽车充电特性的随机特性对电力系统的冲击不断增强,准确及时的源荷预测是实现增强电力系统适应性和稳定性的重要课题。因此,提出一种基于共享权重长短期记忆网络(shared weight long short-term networks,SWLSTM)与St...针对光伏出力和电动汽车充电特性的随机特性对电力系统的冲击不断增强,准确及时的源荷预测是实现增强电力系统适应性和稳定性的重要课题。因此,提出一种基于共享权重长短期记忆网络(shared weight long short-term networks,SWLSTM)与Stacking集成模型相结合的源荷区间预测方法。首先,光伏出力存在时序性特征,采用局部线性嵌入改进k-means算法聚类提取特征日,在实现数据降维同时,减少了网络训练难度;其次,在Stacking集成模型的框架下,将SWLSTM作为元学习器,并通过Q统计量筛选合适的基学习器模型,从而实现多模型融合的多异学习器Stacking集成学习的源荷预测;紧接着,为了得到预测的不确定信息,引入置信度区间预测;最后,采用实测数据对本文所提方法进行验证。结果表明改进k-means算法能够降低其求解难度,加快求解速度,可以快速获取聚类特征;所引入集成学习模型和置信度区间,有效表征源荷预测的不确定性,提升区间预测模型的泛化能力。展开更多
针对现有语音关键词检测方法定位精度低的问题,提出了一种基于多尺度距离矩阵的语音关键词检测与细粒度定位方法(spoken term detection and fine-grained localization method based on multi-scale distance matrices,MF-STD)。该方...针对现有语音关键词检测方法定位精度低的问题,提出了一种基于多尺度距离矩阵的语音关键词检测与细粒度定位方法(spoken term detection and fine-grained localization method based on multi-scale distance matrices,MF-STD)。该方法首先利用残差卷积网络提取特征并构建距离矩阵以建模输入之间的相关性;其次通过多尺度分割和解耦头学习不同尺度下的定位信息;最后根据多尺度加权定位损失、置信度损失和分类损失优化模型,实现对关键词存在性和时域边界的细粒度预测。在LibriSpeech数据集上的实验结果表明,MF-STD在集内词的检测中,精准率和交并比分别达到97.1%和88.6%;在集外词的检测中,精准率和交并比分别达到96.7%和88.2%。与现有的语音关键词检测与定位方法相比,MF-STD的检测准确率和定位精度显著提升,充分证明该方法的先进性,也证明了多尺度特征建模与细粒度定位约束在语音关键词检测任务中的有效性。展开更多
The long-term enhancement in glutamate receptor mediated excitatory responses has been observed in stroke model. This pathological form of plasticity, termed post-ischemic long-term potentiation （i-LTP）, points to f...The long-term enhancement in glutamate receptor mediated excitatory responses has been observed in stroke model. This pathological form of plasticity, termed post-ischemic long-term potentiation （i-LTP）, points to functional reorganization after stroke. Little is known, however, about whether and how this i-LTP would affect subsequent induction of synaptic plasticity. Here, we first directly confirmed that i-LTP was induced in the endothelin-l-induced ischemia model as in other in vitro models. We also demonstrated increased expression of NR2B, CaMKII and p-CaMKII, which are reminiscent of i-LTP. We further induced LTP of field excitatory post- synaptic potentials （fEPSPs） on CA1 hippocampal neurons in peri-infarct regions of the endothelin-l-induced mini-stroke model. We found that LTP of fEPSPs, induced by high-frequency stimulation, displayed a progressive impairment at 12 and 24 hours after ischemia. Moreover, using in vivo multi-channel recording, we found that the local field potential, which represents electrical property of cell ensembles in more restricted regions, was also dam- pened at these two time points. These results suggest that i-LTP elevates the induction threshold of subsequent synap- tic plasticity. Our data helps to deepen the knowledge of meta-synaptic regulation of plasticity after focal ischemia.展开更多
Background and Objective:CybeKnife is a newly developed technology in the field of stereotactic radiosurgery/radiotherapy (SRS/SRT).Compared with conventional SRS/SRT, there are many advantages for CyberKnife in terms...Background and Objective:CybeKnife is a newly developed technology in the field of stereotactic radiosurgery/radiotherapy (SRS/SRT).Compared with conventional SRS/SRT, there are many advantages for CyberKnife in terms of treating tumors that move with respiration, being real-time image-guidance, frameless, high accurateness, and so on.Recently, it has been used to treat different types of malignant carcinoma including intracranial and caudomedial tumors.This study was designed to evaluate the short-term efficacy and toxicity of the CyberKnife radiotherapy for locally advanced pancreatic cancer.Methods:A total of 20 patients with locally advanced (stage II-III) pancreatic cancer treated with CyberKnife were recruited between April 2009 and December 2009.Of 20 patients, 13 were with cancer located at the pancreatic head and 7 were located at the pancreatic body and tail.The planning target volume (PTV) was defined as gross tumor volume (GTV) plus 2-3 mm, and more than 95% PTV should be covered by 75% isodose surface.The median of PTV was 47 cm3 (26-64 cm3).The median total prescription dose was 40 Gy (32-55 Gy) at 3-6 fractions.During treatment delivery, X-Sight Spine Tracking System was used in 5 patients to track movement of the tumor.Other 15 patients were implanted fiducials in the tumors to track movement of the tumor and patient breathing patterns.Results:The median follow-up time was 7 months (3-11 months).All patients had finished the treatment and 19 were alive by the last follow-up.Slight fatigue was the most common complain.Evaluated by CT scan, 6 were complete response, 9 were partial response, 3 were stable disease, and 1 was progression; 1 was dead.There were 6 patients with grade I granulocytopenia, 7 with grade I nausea, and 5 with grade II vomiting.Conclusions:The CyberKnife radiosurgery for the locally advanced pancreatic cancer shows a high rate of local control and minimal toxicity.Long-term follow-up is necessary to evaluate the survival and late toxicity.展开更多
基金the Illinois Department of TransportationAdditional assistance provided by Smart Structures Int
文摘In this paper we present a comparative analysis of global frequency and local deformation data for a large concrete bridge. The asymptotic probability distributions of the central statistics are presented, and compared with empirical bootstrap estimates. Bootstrapped distributions are calculated from reference data obtained during 1999–2000 and used to develop change-point alarm criteria for the structure, using reasonable sensitivity measures developed from FEM simulations and structural analysis. The implications of the frequency data are discussed in conjunction with the strain and displacement measurements in order to discern if the load carrying capacity of the bridge has been affected. The critical need for more advanced temperature compensation models for large structures continually in thermal disequilibrium is discussed.
文摘针对光伏出力和电动汽车充电特性的随机特性对电力系统的冲击不断增强,准确及时的源荷预测是实现增强电力系统适应性和稳定性的重要课题。因此,提出一种基于共享权重长短期记忆网络(shared weight long short-term networks,SWLSTM)与Stacking集成模型相结合的源荷区间预测方法。首先,光伏出力存在时序性特征,采用局部线性嵌入改进k-means算法聚类提取特征日,在实现数据降维同时,减少了网络训练难度;其次,在Stacking集成模型的框架下,将SWLSTM作为元学习器,并通过Q统计量筛选合适的基学习器模型,从而实现多模型融合的多异学习器Stacking集成学习的源荷预测;紧接着,为了得到预测的不确定信息,引入置信度区间预测;最后,采用实测数据对本文所提方法进行验证。结果表明改进k-means算法能够降低其求解难度,加快求解速度,可以快速获取聚类特征;所引入集成学习模型和置信度区间,有效表征源荷预测的不确定性,提升区间预测模型的泛化能力。
文摘针对现有语音关键词检测方法定位精度低的问题,提出了一种基于多尺度距离矩阵的语音关键词检测与细粒度定位方法(spoken term detection and fine-grained localization method based on multi-scale distance matrices,MF-STD)。该方法首先利用残差卷积网络提取特征并构建距离矩阵以建模输入之间的相关性;其次通过多尺度分割和解耦头学习不同尺度下的定位信息;最后根据多尺度加权定位损失、置信度损失和分类损失优化模型,实现对关键词存在性和时域边界的细粒度预测。在LibriSpeech数据集上的实验结果表明,MF-STD在集内词的检测中,精准率和交并比分别达到97.1%和88.6%;在集外词的检测中,精准率和交并比分别达到96.7%和88.2%。与现有的语音关键词检测与定位方法相比,MF-STD的检测准确率和定位精度显著提升,充分证明该方法的先进性,也证明了多尺度特征建模与细粒度定位约束在语音关键词检测任务中的有效性。
基金supported by Major State Basic Research Program of China(Grant No.2013CB733801)
文摘The long-term enhancement in glutamate receptor mediated excitatory responses has been observed in stroke model. This pathological form of plasticity, termed post-ischemic long-term potentiation （i-LTP）, points to functional reorganization after stroke. Little is known, however, about whether and how this i-LTP would affect subsequent induction of synaptic plasticity. Here, we first directly confirmed that i-LTP was induced in the endothelin-l-induced ischemia model as in other in vitro models. We also demonstrated increased expression of NR2B, CaMKII and p-CaMKII, which are reminiscent of i-LTP. We further induced LTP of field excitatory post- synaptic potentials （fEPSPs） on CA1 hippocampal neurons in peri-infarct regions of the endothelin-l-induced mini-stroke model. We found that LTP of fEPSPs, induced by high-frequency stimulation, displayed a progressive impairment at 12 and 24 hours after ischemia. Moreover, using in vivo multi-channel recording, we found that the local field potential, which represents electrical property of cell ensembles in more restricted regions, was also dam- pened at these two time points. These results suggest that i-LTP elevates the induction threshold of subsequent synap- tic plasticity. Our data helps to deepen the knowledge of meta-synaptic regulation of plasticity after focal ischemia.
基金Youth Foundation of Nanjing General Hospital of Nanjing Military Command (No.2009Q051)
文摘Background and Objective:CybeKnife is a newly developed technology in the field of stereotactic radiosurgery/radiotherapy (SRS/SRT).Compared with conventional SRS/SRT, there are many advantages for CyberKnife in terms of treating tumors that move with respiration, being real-time image-guidance, frameless, high accurateness, and so on.Recently, it has been used to treat different types of malignant carcinoma including intracranial and caudomedial tumors.This study was designed to evaluate the short-term efficacy and toxicity of the CyberKnife radiotherapy for locally advanced pancreatic cancer.Methods:A total of 20 patients with locally advanced (stage II-III) pancreatic cancer treated with CyberKnife were recruited between April 2009 and December 2009.Of 20 patients, 13 were with cancer located at the pancreatic head and 7 were located at the pancreatic body and tail.The planning target volume (PTV) was defined as gross tumor volume (GTV) plus 2-3 mm, and more than 95% PTV should be covered by 75% isodose surface.The median of PTV was 47 cm3 (26-64 cm3).The median total prescription dose was 40 Gy (32-55 Gy) at 3-6 fractions.During treatment delivery, X-Sight Spine Tracking System was used in 5 patients to track movement of the tumor.Other 15 patients were implanted fiducials in the tumors to track movement of the tumor and patient breathing patterns.Results:The median follow-up time was 7 months (3-11 months).All patients had finished the treatment and 19 were alive by the last follow-up.Slight fatigue was the most common complain.Evaluated by CT scan, 6 were complete response, 9 were partial response, 3 were stable disease, and 1 was progression; 1 was dead.There were 6 patients with grade I granulocytopenia, 7 with grade I nausea, and 5 with grade II vomiting.Conclusions:The CyberKnife radiosurgery for the locally advanced pancreatic cancer shows a high rate of local control and minimal toxicity.Long-term follow-up is necessary to evaluate the survival and late toxicity.
