This post is from Coursera’s Machine Learning Course (by Andrew Ng) from Stanford Univ.

What is Machine Learning?

ML definition:

1. Arthur Samuel:(older def)Field of study that gives computers the ability to learn without being explicitly programmed

2. Tom Mitchell(Modern Def)Well-posed Learning problem: computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T , as measured by P, improves with experience E

Let’s see this in Playing checkers.

E = the experience of playing many games of checkers
T = the task of playing checkers
P = the probability that the program will win the next game

so, computer program is learn from E to do T
and the program improves with performance measur P

Machine learning algorithms:

there are two types of machine learning algorithms. -> supervised learning -> unsupervised learning

Others : Reinforcement learning, recommender systems

1. Supervised Learning

“right answers”given

we gave a data set of houses
and told it what is the right price.

the task: “just produce more of these right answers”
(x라는 값이 주어지고, y라는 값을 내놓아야 하는게 supervised learning)

1-1) Regression(Linear regression)

• x라는 값에 대해서 y라는 값을 어떤 연속적인 형태로 나타낼 수 있는 것을 Linear regression이라고 표현합니다.

• predict results within a continuous output.
• so, we should map imput variables to some sort of continuous function.

1-2) Classification (Logistic Regression)

we should know that the tumor is malignant or not. So it is the classification. (there can be multiple types)

• trying to predict results in a discrete output.
• map input variables into discrete categories.

2. Unsupervised Learning

just have data with no labels..anything

-> find something in the data 그냥 던져주고 자..cluster해봐

• allows us to approach problems with little or no idea what our results should look like.
• we can derive structures from data where we don’t know the effect of the variables.
• no feedback based on the prediction results.

2-1) Clustering

google news

2-2) Non - Clustering

cocktail party (find structure in a chaotic environment)