ID 原文 译文
57438 通过前向学习的方式从图像中无监督地学得卷积滤波器，在人脸欺诈检测应用场合下，对前向学习网络进行了改进，改进后的网络使用了面向人脸欺诈检测任务的卷积滤波器. The convolutional filters are learned unsupervisedly from the images in a feedforward manner. The feedforward learning network is adapted in the spoof face detection applications by using face anti- spoofing task-oriented convolutional filters learned from the training images.
57439 使用主成分分析变换所得的最小特征值对应的特征向量作为卷积滤波器提取图像的特征. The eigenvectors that corre- spond to the smallest eigenvalues obtained from the principle component analysis transform are used as convolution filters for extracting features from images.
57440 将所提方法在 CASIA-FASD、Idiap Ｒeplay-Attack 和 OULU-NPU 数据集上进行了验证，实验结果表明，在少样本跨攻击类型实验中，所提方法显著提升了欺诈人脸检测的准确率. The method is evaluated on some benchmark data- sets including CASIA-FASD dataset，Idiap Ｒeplay-Attack dataset and OULU-NPU dataset. Experiments show that under the cross presentation attack detection experiments，the proposed method significantly im- proves the classification accuracy of existing methods.
57441 在建模类攻击场景下，基于多元高斯分布的模板攻击是常用的侧信道逆向分析方法. In the profiled scenario，the common method of reverse analysis is the template attack based on multi-Gaussian distribution.
57442 在同样的场景下，分析了深度学习方法在逆向分析领域的应用，提出了基于深度学习的 S 盒逆向分析算法. The article applies the concept of deep learning to the field of reverse a- nalysis for the first time under the same conditions，and proposes an S-box reverse analysis algorithm based on deep learning.
57443 通过选取适用于侧信道逆向分析的深度学习算法、损失函数和标签设计，对类 SM4 算法进行了 S 盒逆向恢复实验. By selecting the deep learning algorithm，loss function and label design method suitable for side channel reverse analysis，an S-box reverse recovery experiment is conducted for SM4-like algorithm.
57444 实验结果表明，使用深度学习进行 S 盒逆向分析是可行的，且在一定的条件下优于模板攻击; It is shown that it is feasible to employ deep learning method to carry out S-box reverse analy- sis，which can have better performance comparing to template attack under certain circumstances.
57445 另外，多层感知机算法预测的结果要优于卷积神经网络算法预测的结果. Moreo- ver，the predicting effect of multi-layer perception algorithm surpasses that of convolutional neural net- work algorithm.
57446 针对现有动态手势识别方法环境适应性低、计算复杂的问题，提出了一种基于视频数据特性的动态手势识别方法. Aiming at the challenges to scene adaptability and computational complexity of dynamic ges- ture recognition，a method based on characteristics of encoded video data is proposed.
57447 使用基于密度的聚类算法 DBSCAN 直接从视频编码数据中的运动矢量提取出运动趋势特征，再通过随机森林分类运动趋势，结合卷积神经网络( CNN) 提取的手型特征识别动态手势. Firstly，density- based spatial clustering of applications with noise is used to extract motion trend features from motion vec- tors. Then，the motion trends are classified by random forest. Finally，combined by the hand shape fea- tures extracted by convolutional neural network( CNN) ，the dynamic gesture is recognized.