[机器学习导论] 绪论

机器学习能做什么

1. Character/Digit Recognition
2. Speech
3. Chess
4. Auto Driving
5. Spam Checking
6. Recommendation
7. Face recognition
8. Find Music

机器学习和人工智能的区别

• 人工智能 (科学, 技术)
  • 研究使计算机来模拟人的某些思维过程和智能行为, 主要包括计算机实现智能的原理, 制造类似于人脑智能的计算机, 是计算机能实现更高层次的应用
• 机器学习
  • 成为解决经验型问题的一种方法, 新的编程范式
  • Definition: A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.
  • machine learning addresses the question of how to build computer programs that improve their performance at some task through experience.
  • 让计算机程序发现数据中的规律, 并根据规律给出预测的一种智能技术

经典的机器学习算法分类

分类(classification)

• 把事物按标准分成一些类别 (离散)
• 例子
  • 垃圾邮件诊断
  • 疾病分析
  • 是否发放信用卡
  • 是否录用

回归(regression)

• 由过去, 现在的数据计算出未来的状态 (连续)
• 例子
  • 预测身高
  • 预测年龄
  • 预测方向盘的旋转角度

聚类(clustering unsupervised learning)

• 没有类别的标准, 按事物间的相似性划分成一些类别

课程内容

• Linear Regression
  • LR with One Variable
  • LR with Multiple Variable
  • Linear Discriminative Analysis
• Logistic Regression and Regularization
• Neural Networks
  • Representation
  • Learning
  • Introduction to Deep Learning
• Support Vector Machine
• Learning from Network : PageRank and Graph Ranking
• Dimensionality Reduction & Concept Learning
  • PCA (Principle Component Analysis)
  • NMF (Non-Negative Matrix Factorization)
• Clustering (Unsupervised Learning)
• Bayes Classification
• Ensemble Learning: Boosting, Bagging and EM
• Evaluation of Machine Learning
• Theory of Statistical Learning

问题驱动试学习

• 根据遇到的问题, 去学习相关知识点
• 用机器学习的思路去学(优化)