Lec_1 Intro

목차

@What is Machine Learning?

-How to use data to learn to make better predictions

1. Recommendation Systems  : suggest to something ( recommended movies..)

2. Spam Detection : by using some words or sentences, we can throw the spam mails.

3. Link Prediction : Like a "Linked In" from one person's neighbors,, Linked person...

-Algorithm behavior changes based on data

*This class: some basic machine learning methods

@Two types of Machine Learning

1. Supervised Learning: Given examples of data and their labels, predict labels of new(unseen) data

2. Unsupervised Learning: Given data, build a model or cluster

*There are other types, but we won't get to it in this class

1. Supervised Learning

Classification:

Given labeled data:    (x_i  , y_i)  ... i=1, ...,n    ( here      x:  feature vector 

          y:  label                      )

*where y is discrete, find a rule to predict y values for unseen x

( Set of input examples (x_i , y_i)  ->  ( classification Algorithm) -> ( Prediction Rule)  ;  on last step,,  (New example x  ->  Label y) <-> (Test Data)

^

l-----  Training Data

*Training and test data must be separate!!!

*Performance Measure:

Accuracy (or fraction of correct answers) on test data

Summary:  

1. Classification: Given labeled data (x_i , y_i)  where y is discrete, predict y values for unseen x

Example 1 : Predict if a new patient has flu or not, based on existing patient data  ,,  What is x and y?

->  Features: Properties of patient     ,,     Label:   Flu   //  No Flu

Example 2 : Which digit in the image?   

->   Label: 0,1,2, ... ,9      ,, What are the features?  ;  Option:  vector of pixel colors ----- pixel and black color filled with that pixel.

A multi class classification problem

*choosing features is non-trivial in real applications.

2. Regression :  when 

x : independent variable             ,,         y : dependent variable           

where y is continuous   , design a rule to predict y values for unseen x

2. Unsupervised Learning

2-1. Clustering

given a set of input objects, group them to clusters by similarity

Example 1 : Cluster videos by people in them

Example 2 : Custer documents by topic     ex)    physics:   gravity, laws of motion, electricity      ,,   Math:  geometry, Algebra

*Features: Words in the document

2-2. Dimensionality Reduction 

Given high dimensional data, find a good low dimensional representation.

Example 1: Images,,     # of pixels = 768,,  So 768-dimensional object 

Can we find a lower dimensional representation?

Total summary;   

1. Supervised Learning : Given examples of data and their labels, predict labels of new ( unseen ) data

Examples: Classification, Regression

2.Unsupervised Learning: Given data, build a model

Examples: Clustering, Dimension Reduction, learning HMMs

* This class we will mostly cover discriminative models 

# https://ucsd.tistory.com/35   (Generative VS Discriminative Models)