ID 原文 译文
22135 该文将核极限学习机与纠错输出编码相结合，提出了一种基于有序编码的核极限学习顺序回归模型。 This paper proposes a new ordered code-based kernel extreme learning ordinal regression machine to fill this gap, which combines the kernel ELM and error correcting output codes effectively.
22136 该模型有效解决了如何在顺序回归中取得良好的特征映射以及如何避免传统极限学习机中隐层节点个数依赖于人工设置的问题。 The model overcomes the problems of how to get high quality feature mappings in ordinal regression and how to avoid setting the number of hidden nodes by manual.
22137 为验证提出模型的有效性，该文在多个顺序回归数据集上进行了测试，测试结果表明，相比于传统 ELM 模型，该文提出的模型在准确率上平均提升了 10.8%，在数据集上预测表现最优，而且获得了最短的训练时间，从而验证了模型的有效性。 To validate the effectiveness of this model, numerous experiments are conducted on a lot of datasets. The experimental results show that the model can improve the accuracy by 10.8% on average compared with traditional ELM-based algorithms and achieve the state- of-the-art performance with the least time.
22138 首先利用判别响应图拟合与 KLT 跟踪算法消除人脸的刚性运动干扰； Firstly, the discriminative response map fitting method and KLT tracking algorithm are used to eliminate the influence of face rigid motion.
22139 该文针对现有的人脸视频心率检测方法在现实情景中受运动干扰难以准确估计心率的问题，提出一种抑制运动干扰的非接触式心率估计新方法。 A novel non-contact heart rate estimation method is proposed to deal with the issue of heart rate measurement from face videos under motion interference in realistic situations, it is hard to estimate heart rate accurately using existing methods.
22140 然后使用对运动鲁棒的色度特征进行两步心率估计，并引入空间梯度因子调控空域和频域的权重，抑制非刚性运动的干扰； Then the chrominance features are selected to estimate heart rate with two steps because of the robustness to facial movements. The frequency and spatial domain weights are assigned through spatial gradient to eliminate the influence of non-rigid motion.
22141 最终得到人脸不同区域融合的平均心率数值和信号波形图，实现心率的精确估计。 Finally, the accurate average heart rate value and pulse wave signal waveform can be acquired from different face regions.
22142 实验结果表明：所提方法相比其它的基于人脸视频的心率估计方法优势明显，提升了信号波形图和真实脉搏波形的一致性，进一步提高了心率估计的精度和鲁棒性。 Compared with three other methods, experimental results indicate that the proposed method enhances the consistency of estimated waveform and ground truth waveform and has obvious superiority in accuracy and robustness of heart rate estimation.
22143 针对卷积神经网络(CNN)在嵌入式端的应用受实时性限制的问题，以及 CNN 卷积计算中存在较大程度的稀疏性的特性，该文提出一种基于 FPGA 的 CNN 加速器实现方法来提高计算速度。 Concerning the problem of real-time restriction on the application of Convolution Neural Network (CNN) in embedded field, and the large degree of sparsity in CNN convolution calculations, this paper proposes an implement method of CNN accelerator based on FPGA to improve computation speed.
22144 首先，挖掘出 CNN 卷积计算的稀疏性特点； Firstly, the sparseness characteristics of CNN convolution calculation are seeked out.