Machine Learning 01

Machine Learning

What is Machine Learning:
- Define a set of funtion
- goodness of function
- pick the best function

---

About Machine Learning

---

Supervised Learning

Regression&classification ∈ Supervised Learning(Learning form teacher).
Supervised Learning needs Training Data.
Supervised Learning is a defined Answer teach.
Supervised Learning needs Input/Output,Pairs of target,function/Output=Label.

Regression

The output of the target function f is scalar.

Classification

Binary Classification

Function f -> Yes/No(Only two answer).
e.g.spam.

Multi-Class Classification

Function f -> class1/class2/class3/……(many classes to choose).
e.g.news on google to classify.

Linear Model

Non-Linear Model

e.g.Deep Learning:Image Recognition,Play go.
SVM.
Decision Tree.
K-NM.

Structured Learning

Beyond Classification.
Output a complex event.
e.g.Face Recognition.

Semi-supervised Learning

Decrease the amount of data.
Only have Labeled and Unlabeled.

Transfer Learning

Labeled&Unlabeled&No Related Labeled&No Related Unlabeled.

Unsupervised Learning

Unlabeled or words learning.
e.g.Machine reading:Machine learns the meaning of words from reading a lot of docs.
Machine drawing：only inputs.

Reinforcement Learning

Learning form critics.
Machine only has a score.