1.1. Motivations

Avandance of data

• text
• image
• timebase
• geo-space
• social network

 

Examples of ML Application

• Document Classification
• Tock Market Prediction
• Spam Filtering and Importance(Tagging)
• SNS Recommendation
• Plate Num Recognition
• Helicopter(Robot) Control
• Opinion Mining, Implicit System

 

Types of Machine Learning

• Supervised Learning : 알아 맞추기, 예측
• Unsupervised Learning : 요약, 정리, 대표, 군집
• Reinforcement Learning : 좀더 우리가 바라는 지능적인 행동(좀더 robotics)

 

Supervised Learning

• You know the true value, label
• You can provide examples of the true value
• 미리 가이드가 정해진 데이터들을 학습
• Classification : true/false
• Regression : Grouping, Ranking, Types, Value prediction
• Case
• Spam filtering
• Automatic grading
• Automatic categorization

 

 

Unsupervised Learning

• You don’t know the true value
• 순수히 기계가 주어진 데이터를 활용해 군집/패턴을 찾을 때
• 있는 데이터를 그대로 분석
• Clustering : estimating sets and affiliations of instances to the sets
• Filtering : estimating underlying and fundamental signals from the mixture of signals and noises
• Case
• Discovering clusters, latent factors, graph structures