ID	原文	译文
53627	根据对数据的参数学习结果，高斯过程混合模型便可自适应地得到每个时刻对应的风险等级，并在预测瓦斯浓度时对各个高斯过程分量的预测进行加权，得到更为鲁棒的预测结果。	According to the MGP with the learned parameters, the risk level at each time can be adaptively computed. As for the gas concentration prediction, we can weight the prediction results of four Gaussian processes together to get a more robust prediction.
53628	实验结果表明，基于高斯过程混合模型的方法可有效地预测瓦斯浓度、评估安全状态。	The experimental results demonstrate that the MGP based method can effectively predict the gas concentration and evaluate the safety state.
53629	针对水声通信系统中低密度校验( Low Density Parity Check，LDPC) 码存在的译码复杂度高和译码错误平层问题，设计了基于极化码编码的水声通信系统，并针对水声信道的极化问题提出了新的基对称扩展极化权重( Polarization Weight，PW) 信道极化法。	Aiming at the problem of high decoding complexity and error floor of LDPC code in underwater acoustic communication system, a polar coded underwater communication system is designed, and a base symmetric spread channel polarization algorithm is proposed for underwater multipath channel polarization.
53630	该算法通过 PW 高阶基计算各个子信道的极化权重，实现了独立于信道状态的信道极化，解决了传统 Polar 码编码稳健性差和依赖信道状态的问题。	By introducing high order base of PW, modified PW solved the problems of poor robustness and channel state dependence of polar code.
53631	理论研究和仿真结果表明，与传统信道极化方法相比，改进的 PW 方法具有更稳定的信道极化结果。	Theoretical analysis and simulation results show that the modified PW method has stronger robustness in channel polarization than traditional polarization meth- od.
53632	将该方法应用于正交频分复用( Orthogonal Fre- quency Division Multiplexing，OFDM) 水声通信系统，与现有的 LDPC 编码方法相比，基于改进的 PW 极化码具有更低的通信误码率和译码复杂度，且不存在译码错误平层。	The proposed method is applied to OFDM underwater acoustic communication system, the designed Polar-OFDM system improves the performance in terms of the bit error rate ( BEＲ) and decoding complexity sufficiently in comparison with LD- PC, and without error floor.
53633	随着深度学习技术的迅猛发展，各种相似的骨干网络被用于多源遥感分类任务中，取得了很高的识别正确率。	With the rapid development of deep learning, the same set of backbone network architectures have often been adopted for multi-source remote sensing image classification and achieved impressive results.
53634	然而，由于神经网络极易受到对抗样本的攻击，这给遥感任务带来了很高的安全隐患。	However, this raises severe security issues for remote sensing task as these models are known to be vulnerable to adversarial attacks.
53635	以往的对抗攻击方法可有效攻击单波段遥感图像的分类器，但不同波段的攻击并不耦合，这导致现有方法在现实世界中难以用于多源分类器的攻击。	Conventional adversarial attack methods are able to fool single-band images classifier, but the adversarial perturbations of different band are not coupling, that results in its unavailability in multi-source remote sensing images attack.
53636	针对多源遥感的特性，本文提出了一种新的基于稀疏差分协同进化的对抗攻击方法:	In this paper, we propose an adversarial attack method with consistent perturbation pattern across multi-source images, using sparse differential coevo- lution: