PyTorch 深度学习实践（刘二大人）---第一讲Overview

算法：穷举法、贪心法、分治法、动态规划

AI>机器学习（如：逻辑回归）>表征学习（如：浅层自动编码器）>深度学习（如：神经网络）

Rule-base system: input->hand_designed->output

Classic machine-learning: input->hand_designed features->mapping from features->output

Representation learning: input->features->mapping from features->output

Deep learning(End2End): input->simple features->additional layers of more abstract features->mapping from features->output

维度灾难

 

present 流形manifold(降维)

Scikit-learn: 分类、聚类、回归、降维

Back Propagation: 链式法则