Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices...Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.展开更多
目的评价缺铁性贫血(IDA)和β地中海贫血(β-TT)的鉴别诊断公式在四川自贡地区人群中的应用价值。方法选取760例来自四川自贡地区患者,其中β-TT 255例、IDA505例,应用红细胞(RBC)参数计算出10个鉴别公式(分别为Shine and Lal公式、Gree...目的评价缺铁性贫血(IDA)和β地中海贫血(β-TT)的鉴别诊断公式在四川自贡地区人群中的应用价值。方法选取760例来自四川自贡地区患者,其中β-TT 255例、IDA505例,应用红细胞(RBC)参数计算出10个鉴别公式(分别为Shine and Lal公式、Green and King公式、Mentzer公式、RDWI公式、Srivastava公式、Ricerca公式、England and Fraser公式、Ehsani公式、Sirdah公式和Huber-Herklotz公式)的敏感度(SEN)、特异度(SPE)、阳性预测值(PPV)、阴性预测值(NPV)、约登指数(YI),并绘制受试者工作特征(ROC)曲线。结果 Mentzer公式的诊断价值最高,其YI最高(80.1%),诊断IDA的SEN为97.0%,SPE为83.1%,曲线下面积(AUC)最高(0.971);其次为RDWI公式,YI、SEN、SPE、AUC分别为72.8%、95.2%、77.6%、0.961。结论在四川自贡地区人群中Mentzer公式和RDWI公式具有较好的鉴别诊断IDA和β-TT的能力。展开更多
CO2 geological sequestration in a depleted shale gas reservoir is a promising method to address the global energy crisis as well as to reduce greenhouse gas emissions. Though improvements have been achieved by many re...CO2 geological sequestration in a depleted shale gas reservoir is a promising method to address the global energy crisis as well as to reduce greenhouse gas emissions. Though improvements have been achieved by many researchers, the carbon sequestration and enhanced gas recovery(CS-EGR) in shale formations is still in a preliminary stage. The current research status of CO2 sequestration in shale gas reservoirs with potential EGR is systematically and critically addressed in the paper. In addition, some original findings are also presented in this paper. This paper will shed light on the technology development that addresses the dual problem of energy crisis and environmental degradation.展开更多
Locating wells is an important step in oil exploitation. This paper proposes a novel approach, which first combines particle swarm optimization, genetic algorithm, and a reservoir simulation evaluation tool to optimiz...Locating wells is an important step in oil exploitation. This paper proposes a novel approach, which first combines particle swarm optimization, genetic algorithm, and a reservoir simulation evaluation tool to optimize the locations of vertical wells. Simulation results show that the convergence efficiency of our approach outperforms traditional genetic algorithm and overcomes the disadvantage of particle swarm algorithm that would be easily trapped into best-at-local solution so that its optimization result has been significantly improved.展开更多
基金supported by the National Natural Science Foundation of China(62171088,U19A2052,62020106011)the Medico-Engineering Cooperation Funds from University of Electronic Science and Technology of China(ZYGX2021YGLH215,ZYGX2022YGRH005)。
文摘Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.
文摘目的评价缺铁性贫血(IDA)和β地中海贫血(β-TT)的鉴别诊断公式在四川自贡地区人群中的应用价值。方法选取760例来自四川自贡地区患者,其中β-TT 255例、IDA505例,应用红细胞(RBC)参数计算出10个鉴别公式(分别为Shine and Lal公式、Green and King公式、Mentzer公式、RDWI公式、Srivastava公式、Ricerca公式、England and Fraser公式、Ehsani公式、Sirdah公式和Huber-Herklotz公式)的敏感度(SEN)、特异度(SPE)、阳性预测值(PPV)、阴性预测值(NPV)、约登指数(YI),并绘制受试者工作特征(ROC)曲线。结果 Mentzer公式的诊断价值最高,其YI最高(80.1%),诊断IDA的SEN为97.0%,SPE为83.1%,曲线下面积(AUC)最高(0.971);其次为RDWI公式,YI、SEN、SPE、AUC分别为72.8%、95.2%、77.6%、0.961。结论在四川自贡地区人群中Mentzer公式和RDWI公式具有较好的鉴别诊断IDA和β-TT的能力。
基金supported by the General Project of National Natural Science Foundation of China (Grant Nos. 51974253 and 51974247)the Youth Project of National Natural Science Foundation of China (Grant No.41502311)+1 种基金the Natural Science Foundation of Shaanxi Province (Grant No.2019JQ-525)the Natural Science Basic Research Program of Shaanxi Province (Grant No. 2020JQ-781)。
文摘CO2 geological sequestration in a depleted shale gas reservoir is a promising method to address the global energy crisis as well as to reduce greenhouse gas emissions. Though improvements have been achieved by many researchers, the carbon sequestration and enhanced gas recovery(CS-EGR) in shale formations is still in a preliminary stage. The current research status of CO2 sequestration in shale gas reservoirs with potential EGR is systematically and critically addressed in the paper. In addition, some original findings are also presented in this paper. This paper will shed light on the technology development that addresses the dual problem of energy crisis and environmental degradation.
基金Supported by the National Natural Science Foundation of China (61070008)the Jiangxi Province Science and Technology Pillar Program (2009BHB16400)
文摘Locating wells is an important step in oil exploitation. This paper proposes a novel approach, which first combines particle swarm optimization, genetic algorithm, and a reservoir simulation evaluation tool to optimize the locations of vertical wells. Simulation results show that the convergence efficiency of our approach outperforms traditional genetic algorithm and overcomes the disadvantage of particle swarm algorithm that would be easily trapped into best-at-local solution so that its optimization result has been significantly improved.
