What is machine learning?
Using computers to find out the rules of historical data and using them to make decisions about uncertain future scenarios
Learning mode?
Offline learning
Online learning
Typical application of machine learning?
Association rules: "Beer + diaper"
Cluster: Global (for frequent global travel people), dynamic zone (for students with more traffic in school), Shenzhou (for workers, white-collar often make long-distance calls)
Naive Bayes: Junk mail
Decision Tree: Credit card fraud
CTR estimate: Internet Advertising
Collaborative filtering: Recommended systems
Natural Speech processing: Emotion Analysis and entity recognition
Deep Learning: Image recognition
More applications:
Speech recognition
Personalised Medical
Sentiment analysis
Human Face recognition
Automatic driving
Smart Robot
Private Virtual Assistant
Gesture Control
Automatic video content recognition
Real-time Machine translation
Data features
Transaction data (using relational data) vs behavioral data (development of NOSQL data)
Small amount of data vs massive data
Mining volume analysis vs full-scale analysis
Data analysis (Reporting on past events), analytical methods (user-driven, interactive analysis), participants (data analyst: Data + algorithms)
Machine learning (predicting future events), analytical methods (data driven, automatic knowledge discovery), participants (algorithms)
Algorithm Classification (1):
Supervised learning: Classification algorithm, regression algorithm
Unsupervised Learning: Clustering
Semi-supervised learning: children walking
Algorithm Classification (2):
Classification and regression
Clustering
Marking
Algorithm Classification (3):
Build model
discriminant mode
Getting Started with machine learning