ID	原文	译文
53937	首先，进行图像预处理，包括统一尺寸、人脸对齐、图像增强、归一化和图像剪裁等;	Firstly, we perform image preprocessing, including uniform size, face alignment, image enhancement, normalization, and image cropping.
53938	其次，初始化构建的改进十字绣网络，并将层与层之间的共享称之为“微共享”，将模块与模块之间的共享称之为“模块共享”;	Secondly, we initialize the constructed Improved Cross-Stitch Network, in which the sharing between layers is called “Micro-sharing”, and the sharing between modules is called “Module sharing”.
53939	最后，对训练模型进行测试。	Finally, the trained model is tested.
53940	实验结果表明，采用改进十字绣网络，人脸美丽预测取得63. 95%的准确率，高于常规方法最高准确率;	Experimental results show that by Improved Cross-Stitch Network, the accuracy of Facial Beauty Prediction is 63. 95%, higher than the highest accuracy rate by conventional methods.
53941	为多任务学习提供了一种新思路。	Therefore, the Improved Cross-Stitch Network provides a new idea for Multi-task Learning.
53942	NOMA(Non-orthogonal Multiple Access，NOMA)系统中的用户分簇策略对系统性能有着极大的影响。	The user clustering strategy in NOMA(Non-orthogonal Multiple Access, NOMA) system has a great impact on system performance.
53943	该文主要研究NOMA下行链路的用户动态分簇问题，其目的是最大化系统总吞吐量。	This paper mainly studied the user clustering problem of NOMA downlink, and its main purpose was to maximize the total system throughput.
53944	与大多数文章不同，该研究对簇中用户数以及簇个数都没有限制。	The difference from most previous articles was that this study did not limit the number of users in a cluster and the number of clusters.
53945	遗传算法可用于优化NOMA网络中的用户动态分簇，但标准遗传算法存在收敛速度慢且容易陷入局部最优的问题。	The standard genetic algorithm can be used to optimize the dynamic clustering of users in the NOMA network, but it has the problem of slow convergence and easy to fall into local optimum.
53946	基于此，该文将自适应调节参数的改进遗传算法用于用户的动态分簇，来改善上述问题。	Based on this, this paper used an improved genetic algorithm with adaptive adjustment parameters for dynamic clustering of users to improve the above problems.