中文摘要：

提出一种改进的遗传神经网络:将遗传算法与BP神经网络相结合,学习过程中加入了LM(Levenberg-Marquadt)算法.利用遗传算法随机提取训练集的视频特征,再用提取特征之后的训练集训练该网络,通过LM对网络进行精调,从而得到必要的特征子集优化网络结构.将该网络应用到视频特征提取及篡改视频分类,实验结果表明,该网络能够快速有效的筛选出显著特征,得到最优特征子集,实现对是否为篡改视频进行分类的目的.

英文摘要：

An improved genetic neural network is proposed in this paper:combined genetic algorithm and BP neural network,Levenberg-Marquadt algorithm is added to the learning process.The video features of the training set are randomly extracted by genetic algorithm,then the network is trained by the training set which is used to extract fea-tures,and the network is refined by LM,the feature subset is obtained so as to optimize the network structure.The net-work is applied to video feature extraction and tampering video classification,experimental results show that the net-work can effectively filter out the salient feature and get optimal feature subset,it is possible to quickly classify of tam-pering videos.