The neutron flux monitor(NFM)system is an important diagnostic subsystem introduced by large nuclear fusion devices such as international thermonuclear experimental reactor(ITER),Japan torus-60,tokamak fusion test rea...The neutron flux monitor(NFM)system is an important diagnostic subsystem introduced by large nuclear fusion devices such as international thermonuclear experimental reactor(ITER),Japan torus-60,tokamak fusion test reactor,and HL-2 A.Neutron fluxes can provide real-time parameters for nuclear fusion,including neutron source intensity and fusion power.Corresponding to different nuclear reaction periods,neutron fluxes span over seven decades,thereby requiring electronic devices to operate in counting and Campbelling modes simultaneously.Therefore,it is crucial to design a real-time NFM system to encompass such a wide dynamic range.In this study,a high-precision NFM system with a wide measurement range of neutron flux is implemented using realtime multipoint linear calibration.It can automatically switch between counting and Campbelling modes with variations in the neutron flux.We established a testing platform to verify the feasibility of the NFM system,which can output the simulated neutron signal using an arbitrary waveform generator.Meanwhile,the accurate calibration interval of the Campbelling mode is defined well.Based on the above-mentioned design,the system satisfies the requirements,offering a dynamic range of 10~8 cps,temporal resolution of 1 ms,and maximal relative error of 4%measured at the signal-to-noise ratio of 15.8 dB.Additionally,the NFM system is verified in a field experiment involving HL-2 A,and the measured neutron flux is consistent with the results.展开更多
In a crowd density estimation dataset,the annotation of crowd locations is an extremely laborious task,and they are not taken into the evaluation metrics.In this paper,we aim to reduce the annotation cost of crowd dat...In a crowd density estimation dataset,the annotation of crowd locations is an extremely laborious task,and they are not taken into the evaluation metrics.In this paper,we aim to reduce the annotation cost of crowd datasets,and propose a crowd density estimation method based on weakly-supervised learning,in the absence of crowd position supervision information,which directly reduces the number of crowds by using the number of pedestrians in the image as the supervised information.For this purpose,we design a new training method,which exploits the correlation between global and local image features by incremental learning to train the network.Specifically,we design a parent-child network(PC-Net)focusing on the global and local image respectively,and propose a linear feature calibration structure to train the PC-Net simultaneously,and the child network learns feature transfer factors and feature bias weights,and uses the transfer factors and bias weights to linearly feature calibrate the features extracted from the Parent network,to improve the convergence of the network by using local features hidden in the crowd images.In addition,we use the pyramid vision transformer as the backbone of the PC-Net to extract crowd features at different levels,and design a global-local feature loss function(L2).We combine it with a crowd counting loss(LC)to enhance the sensitivity of the network to crowd features during the training process,which effectively improves the accuracy of crowd density estimation.The experimental results show that the PC-Net significantly reduces the gap between fullysupervised and weakly-supervised crowd density estimation,and outperforms the comparison methods on five datasets of Shanghai Tech Part A,ShanghaiTech Part B,UCF_CC_50,UCF_QNRF and JHU-CROWD++.展开更多
Deep learning has recently gained significant prominence in various real-world applications such as image recognition,natural language processing,and autonomous vehicles.While deep neural networks appear to have diffe...Deep learning has recently gained significant prominence in various real-world applications such as image recognition,natural language processing,and autonomous vehicles.While deep neural networks appear to have different architectures,the main operations within these models are matrix-vector multiplications(MVM).Compute-in-memory(CIM)architectures are promising solutions for accelerating the massive MVM operations by alleviating the frequent data movement issue in traditional processors.Analog CIM macros leverage current-accumulating or charge-sharing mechanisms to perform multiply-and-add(MAC)computations.Even though they can achieve high throughput and efficiency,the computing accuracy is sacrificed due to the analog nonidealities.To ensure precise MAC calculations,it is crucial to analyze the sources of nonidealities and identify their impacts,along with corresponding solutions.In this paper,comprehensive linearity analysis and dedicated calibration methods for charge domain static-random access memory(SRAM)based in-memory computing circuits are proposed.We analyze nonidealities from three areas based on the mechanism of charge domain computing:charge injection effect,temperature variations,and ADC reference voltage mismatch.By designing a 256×256 CIM macro and conducting investigations via post-layout simulation,we conclude that these nonidealities don’t deteriorate the computing linearity,but only cause the scaling and bias drift.To mitigate the scaling and bias drift identified,we propose three calibration methods ranging from the circuit level to the algorithm level,all of which exhibit promising results.The comprehensive analysis and calibration methods can assist in designing CIM macros with more accurate MAC computations,thereby supporting more robust deep learning inference.展开更多
Experimental studies on the basic characteristics of IPs applied in T-ray imaging are carried out by utilizing isotopic y-ray sources. The 1.25 MeV T-ray sensitivity of the BAS-MS and BAS-TR imaging plates and their e...Experimental studies on the basic characteristics of IPs applied in T-ray imaging are carried out by utilizing isotopic y-ray sources. The 1.25 MeV T-ray sensitivity of the BAS-MS and BAS-TR imaging plates and their enhanced sensitivity by covering appropriate Compton conversion foils are measured based on the studies of the image intensity linear calibration, time attenuation laws and the influence of scanning parameter settings. The energy-dependent T-ray sensitivity of the IPs is also obtained by the studies of the measured sensitivity and the Monte Carlo simulated energy deposition in the IPs＇ sensitive layer. Furthermore, a method of a sandwich detection structure as well as its primary experimental validations are presented in order to increase the gamma-to-neutron ratio in a y/n mixed radiation field.展开更多
基金supported by the National Natural Science Foundation of China(Nos.11475131,11975307,and 11575184)the National Magnetic Confinement Fusion Energy Development Research(No.2013GB104003)。
文摘The neutron flux monitor(NFM)system is an important diagnostic subsystem introduced by large nuclear fusion devices such as international thermonuclear experimental reactor(ITER),Japan torus-60,tokamak fusion test reactor,and HL-2 A.Neutron fluxes can provide real-time parameters for nuclear fusion,including neutron source intensity and fusion power.Corresponding to different nuclear reaction periods,neutron fluxes span over seven decades,thereby requiring electronic devices to operate in counting and Campbelling modes simultaneously.Therefore,it is crucial to design a real-time NFM system to encompass such a wide dynamic range.In this study,a high-precision NFM system with a wide measurement range of neutron flux is implemented using realtime multipoint linear calibration.It can automatically switch between counting and Campbelling modes with variations in the neutron flux.We established a testing platform to verify the feasibility of the NFM system,which can output the simulated neutron signal using an arbitrary waveform generator.Meanwhile,the accurate calibration interval of the Campbelling mode is defined well.Based on the above-mentioned design,the system satisfies the requirements,offering a dynamic range of 10~8 cps,temporal resolution of 1 ms,and maximal relative error of 4%measured at the signal-to-noise ratio of 15.8 dB.Additionally,the NFM system is verified in a field experiment involving HL-2 A,and the measured neutron flux is consistent with the results.
基金the Humanities and Social Science Fund of the Ministry of Education of China(21YJAZH077)。
文摘In a crowd density estimation dataset,the annotation of crowd locations is an extremely laborious task,and they are not taken into the evaluation metrics.In this paper,we aim to reduce the annotation cost of crowd datasets,and propose a crowd density estimation method based on weakly-supervised learning,in the absence of crowd position supervision information,which directly reduces the number of crowds by using the number of pedestrians in the image as the supervised information.For this purpose,we design a new training method,which exploits the correlation between global and local image features by incremental learning to train the network.Specifically,we design a parent-child network(PC-Net)focusing on the global and local image respectively,and propose a linear feature calibration structure to train the PC-Net simultaneously,and the child network learns feature transfer factors and feature bias weights,and uses the transfer factors and bias weights to linearly feature calibrate the features extracted from the Parent network,to improve the convergence of the network by using local features hidden in the crowd images.In addition,we use the pyramid vision transformer as the backbone of the PC-Net to extract crowd features at different levels,and design a global-local feature loss function(L2).We combine it with a crowd counting loss(LC)to enhance the sensitivity of the network to crowd features during the training process,which effectively improves the accuracy of crowd density estimation.The experimental results show that the PC-Net significantly reduces the gap between fullysupervised and weakly-supervised crowd density estimation,and outperforms the comparison methods on five datasets of Shanghai Tech Part A,ShanghaiTech Part B,UCF_CC_50,UCF_QNRF and JHU-CROWD++.
基金supported in part by the National Key Research and Development Program of China under Grant 2022YFB4400900in part by the Natural Science Foundation of China under Grant 62371223.
文摘Deep learning has recently gained significant prominence in various real-world applications such as image recognition,natural language processing,and autonomous vehicles.While deep neural networks appear to have different architectures,the main operations within these models are matrix-vector multiplications(MVM).Compute-in-memory(CIM)architectures are promising solutions for accelerating the massive MVM operations by alleviating the frequent data movement issue in traditional processors.Analog CIM macros leverage current-accumulating or charge-sharing mechanisms to perform multiply-and-add(MAC)computations.Even though they can achieve high throughput and efficiency,the computing accuracy is sacrificed due to the analog nonidealities.To ensure precise MAC calculations,it is crucial to analyze the sources of nonidealities and identify their impacts,along with corresponding solutions.In this paper,comprehensive linearity analysis and dedicated calibration methods for charge domain static-random access memory(SRAM)based in-memory computing circuits are proposed.We analyze nonidealities from three areas based on the mechanism of charge domain computing:charge injection effect,temperature variations,and ADC reference voltage mismatch.By designing a 256×256 CIM macro and conducting investigations via post-layout simulation,we conclude that these nonidealities don’t deteriorate the computing linearity,but only cause the scaling and bias drift.To mitigate the scaling and bias drift identified,we propose three calibration methods ranging from the circuit level to the algorithm level,all of which exhibit promising results.The comprehensive analysis and calibration methods can assist in designing CIM macros with more accurate MAC computations,thereby supporting more robust deep learning inference.
基金Supported by Youth Foundation of NNSF(11005095)Science and Technology Development Foundation of China Academy of Engineering Physics(2011B0103017)
文摘Experimental studies on the basic characteristics of IPs applied in T-ray imaging are carried out by utilizing isotopic y-ray sources. The 1.25 MeV T-ray sensitivity of the BAS-MS and BAS-TR imaging plates and their enhanced sensitivity by covering appropriate Compton conversion foils are measured based on the studies of the image intensity linear calibration, time attenuation laws and the influence of scanning parameter settings. The energy-dependent T-ray sensitivity of the IPs is also obtained by the studies of the measured sensitivity and the Monte Carlo simulated energy deposition in the IPs＇ sensitive layer. Furthermore, a method of a sandwich detection structure as well as its primary experimental validations are presented in order to increase the gamma-to-neutron ratio in a y/n mixed radiation field.
