The Macao Science Satellite-1 is a two-satellite constellation specifically designed to study the geomagnetic field and particle radiation environment in low Earth orbit,particularly in the South Atlantic Anomaly regi...The Macao Science Satellite-1 is a two-satellite constellation specifically designed to study the geomagnetic field and particle radiation environment in low Earth orbit,particularly in the South Atlantic Anomaly region,with a low inclination orbit.Each of the two MSS-1 satellites carries a medium-energy electron spectrometer(MES).The MES sensor heads are based on pinhole imaging technology,which can simultaneously measure 50-600 keV electrons incident from nine directions with a field of view(FOV)of 180°×30°.The two MESs can realize the pitch angle coverage of medium energy electrons at most positions in the orbit.The MSS-1 A/B MESs can realize direct observation of precipitating electrons and electrons near their loss cones.It can help to study the electron generation mechanism in the inner radiation belt and quantify the precipitation of magnetospheric energetic electrons.Combined with the geomagnetic index,solar wind parameters,interplanetary magnetic field conditions,etc.,it can also help to build a dynamic evolution model of energetic electrons in the near-Earth space,to realize the early warning and prediction of space weather based on the observation data,which can provide safety for spacecraft and astronauts in the nearEarth space.展开更多
In this paper,the L_(2,∞)normalization of the weight matrices is used to enhance the robustness and accuracy of the deep neural network(DNN)with Relu as activation functions.It is shown that the L_(2,∞)normalization...In this paper,the L_(2,∞)normalization of the weight matrices is used to enhance the robustness and accuracy of the deep neural network(DNN)with Relu as activation functions.It is shown that the L_(2,∞)normalization leads to large dihedral angles between two adjacent faces of the DNN function graph and hence smoother DNN functions,which reduces over-fitting of the DNN.A global measure is proposed for the robustness of a classification DNN,which is the average radius of the maximal robust spheres with the training samples as centers.A lower bound for the robustness measure in terms of the L_(2,∞)norm is given.Finally,an upper bound for the Rademacher complexity of DNNs with L_(2,∞)normalization is given.An algorithm is given to train DNNs with the L_(2,∞)normalization and numerical experimental results are used to show that the L_(2,∞)normalization is effective in terms of improving the robustness and accuracy.展开更多
基金supported by the National Natural Science Foundation of China(Grant No.42274225)the Science and Technology Development Fund,Macao SAR(Grant No.SKL-LPS(MUST)-2024-2026)。
文摘The Macao Science Satellite-1 is a two-satellite constellation specifically designed to study the geomagnetic field and particle radiation environment in low Earth orbit,particularly in the South Atlantic Anomaly region,with a low inclination orbit.Each of the two MSS-1 satellites carries a medium-energy electron spectrometer(MES).The MES sensor heads are based on pinhole imaging technology,which can simultaneously measure 50-600 keV electrons incident from nine directions with a field of view(FOV)of 180°×30°.The two MESs can realize the pitch angle coverage of medium energy electrons at most positions in the orbit.The MSS-1 A/B MESs can realize direct observation of precipitating electrons and electrons near their loss cones.It can help to study the electron generation mechanism in the inner radiation belt and quantify the precipitation of magnetospheric energetic electrons.Combined with the geomagnetic index,solar wind parameters,interplanetary magnetic field conditions,etc.,it can also help to build a dynamic evolution model of energetic electrons in the near-Earth space,to realize the early warning and prediction of space weather based on the observation data,which can provide safety for spacecraft and astronauts in the nearEarth space.
基金partially supported by NKRDP under Grant No.2018YFA0704705the National Natural Science Foundation of China under Grant No.12288201.
文摘In this paper,the L_(2,∞)normalization of the weight matrices is used to enhance the robustness and accuracy of the deep neural network(DNN)with Relu as activation functions.It is shown that the L_(2,∞)normalization leads to large dihedral angles between two adjacent faces of the DNN function graph and hence smoother DNN functions,which reduces over-fitting of the DNN.A global measure is proposed for the robustness of a classification DNN,which is the average radius of the maximal robust spheres with the training samples as centers.A lower bound for the robustness measure in terms of the L_(2,∞)norm is given.Finally,an upper bound for the Rademacher complexity of DNNs with L_(2,∞)normalization is given.An algorithm is given to train DNNs with the L_(2,∞)normalization and numerical experimental results are used to show that the L_(2,∞)normalization is effective in terms of improving the robustness and accuracy.
