libsvm工具箱的主要函数为svmtrain和svmpredict，其中一些参数设置可以按照svm的类型和核函数所支持的参数进行任意组合，如果设置的参数在函数或svm类型中没有也不会产生影响，程序不会接受该参数，程序主要用一个例子来说明主要参数的使用。(The main functions of the libsvm toolbox are svmtrain and svmpredict, some of which can be arbitrarily combined according

本示例展示了怎样从一个预处理的卷积神经网络中提取特征，并用这些特征去训练一个图像分类器。(This example shows how to extract learned features from a pretrained convolutional neural network, and use those features to train an image classifier. Feature extraction is the easiest and fastest way use