Stanford Machine Learning Study 2016/7/4

An introductory tutorial on machine learning with a higher degree of identity, by Andrew Ng of Stanford. NetEase public class with Chinese and English subtitles teaching video resources (http://open.163.com/special/opencourse/ machinelearning.html), handout stamp here: http://cs229.stanford.edu/materials.html

There are a variety of similar course learning notes on the Web, which will also be part of my study. Be patient and be curious~

The first section is about machine learning motivation, a brief introduction to supervised learning (supervised learning), unsupervised learning (unsupervised learning) and reinforcement learning (reinforcement learning), It also shows some of the projects that have been completed using machine learning technology demo.

After all, is all-round popular science over deep learning, for the basic concept of a certain understanding, no longer repeat ~

Supervised learning can be considered as a given set of data, is Groundturth, that is (x, y) such (input feature,output) group, the train when the output is taken groundtruth. SVM can be used to realize the mapping of low dimensional linear non-feature to high dimensional linear fractal feature space, and better classification.

Unsupervised learning is not groundtruth, and you can use clustering to explore hidden structural features in your data.

Reinforcement learning can be seen as a feedback environment, for each action it makes, will be the rest of the reward or punishment, the agent to learn from the choice of strategy to make agent performance optimal.

Stanford Machine Learning Study 2016/7/4