• The general steps of Machine Learning

1, Traning Data & Test Data

This step is the most important part in machine learning. Very easy to ignore and underestimate.

Data is wrong, all the steps afterward are wrong.

More than that, it’s very difficult to understand where is wrong, how to impove that.

Maybe the data can be produced by online logs by user.

Maybe the data can not be gotten from machine, only by humans.

2, Feature Selection

The less feature, the faster the model is calculated and updated

The more feature’s meaning is ,the less prabability overfit happens

L1-norm

http://www.cnblogs.com/heaad/archive/2011/01/02/1924088.html

3, Model Choosing

Different model has its personality

4, Evaluation

RSE

FRTP…

Reference:

http://zhfuzh.blog.163.com/blog/static/14553938720127309539465/

http://blog.csdn.net/pennyliang/article/details/6998517

http://deeplearning.stanford.edu/wiki/index.php/UFLDL%E6%95%99%E7%A8%8B

It’s not finished…