ID 原文 译文
22145 其次，为了用好参数稀疏性，把 CNN 卷积计算转换为矩阵相乘； Secondly, in order to use the parameters sparseness, CNN convolution calculations are converted to matrix multiplication.
22146 最后，提出基于 FPGA 的并行矩阵乘法器的实现方案。 Finally, the implementation method of parallel matrix multiplier based on FPGA is proposed.
22147 在 Virtex-7 VC707 FPGA 上的仿真结果表明，相比于传统的 CNN 加速器，该设计缩短了 19%的计算时间。 Simulation results on the Virtex-7 VC707 FPGA show that the design shortens the calculation time by 19% compared to the traditional CNN accelerator.
22148 通过稀疏性来简化 CNN 计算过程的方式，不仅能在 FPGA 实现，也能迁移到其他嵌入式端。 The method of simplifying the CNN calculation process by sparseness not only can be implemented on FPGA, but also can migrate to other embedded ends.
22149 基于平衡态非相干散射理论的 GUISDAP 程序包在拟合含有纯增长模式线的非相干散射功率谱及反演加热扰动电离层参量中存在严重误差，该文提出一种以未加热时刻的电离层参量为先验信息，利用最小二乘法搜寻最佳高斯峰描述纯增长模式线来修正实测非相干散射谱，并使用电子超高斯分布的非相干散射理论拟合修正后非相干散射谱来反演获取加热扰动电子温度的方法。 In the fitting process of the incoherent scatter power spectra with pure-growth mode line and the inversion of the disturbed ionospheric parameters, the GUISDAP package which is based on the equilibrium incoherent scatter theory, always exits serious error. Assuming the ionospheric parameters at heating-off time as a priori information, by using the least squares method to search the best Gaussian peak for the modification of the measured incoherent scatter spectra, one method is provided to obtain the disturbed electron temperature through fitting the modified measured spectra by taking advantage of the incoherent scatter theory of electron super-Gaussian distribution.
22150 通过将该方法用于分析 2010 年秋季我国在挪威开展的电离层加热实验数据表明：利用该方法对含纯增长模式线的实测谱修正拟合后获得的电子温度比未加热时刻提高约 800 K 左右，增幅比例约 13%~50%，该结果与现有文献报道的电离层加热引起的电子温度增幅范围吻合较好。 This method is used to the ionospheric heating experimental data which is conducted in Norway in the autumn of 2010, and the results indicate that the electron temperature obtained by fitting the modified spectra is about 800 K higher than that of the heating off time, and the increase ratio is about 13%~50%, which is well coincided with the electron temperature increase range of ionospheric heating reported in the existing literature.
22151 结论表明该方法适用于基于含纯增长模式线的非相干散射谱反演扰动电离层参量。 The conclusion shows that the method is applicable to the inversion of disturbed ionospheric parameters by using of the incoherent scatter spectra with pure-growth mode line.
22152 对高维数据进行建模是 Takagi-Sugeno(T-S)模糊系统建模面临的一个重大挑战。 It is a great challenge to model Takagi-Sugeno(T-S) fuzzy systems on high dimensional data due to the problem of "the curse of dimensionality".
22153 为此，该文提出一种特征选择与组稀疏编码相结合的模糊系统建模新方法 WOMP-GS-FIS。 To this end, a novel T-S fuzzy system modeling method called WOMP-GS-FIS is proposed. The proposed method considers feature selection and group sparse coding simultaneously.
22154 首先，运用一种新型的加权正交匹配追踪算法对原始样本进行特征选择，在此基础上提取出模糊规则前件并产生模糊系统字典； Specifically, feature selection is performed by a novel Weighted Orthogonal Matching Pursuit (WOMP) method, based on which the fuzzy rule antecedent part is extracted and the dictionary of the fuzzy system is generated.