机器学习定义:A computer program is said to learn from experience E with respect to some classes of task T and performance measure P if its performance can improve with E on T measured by P.
小样本学习定义:Few-Shot Learning (FSL) is a type of machine learning problems (specified by E, T and P), where E contains only a limited number of examples with supervised information for the target T.
1.多任务学习(同时进行多个相关任务训练,共享表示,以获得更好的泛化能力)与迁移学习不同(将源任务中学到的知识运用到目标任务中) parameter sharing:多任务间共享参数(例如最开始几层网络结构共享,最后输出层单独训练) parameter typing:对不同任务的参数正则化处理,使其参数相似(encourages parameters of different tasks to be similar using regularization)
Parameters of Inference Networks:找到最佳的\theta,使得最大化p(z|x;\theta,\gamma)=\frac{p(x,z;\theta,\gamma)}{p(x;\gamma)}=\frac{p(x|z;\theta)p(z;\gamma)}{\int p(x|z;\theta)p(z;\gamma)dz},通常使用从数据中学到的变分分布q(z;\delta)来估计p(z|x;\theta,\gamma)。(?)