- Supervised Learning
- Unsupervised Learning
- Reinforced Learning
Goal：
- How to apply these methods
- How to evaluate each methods
What is Machine Learning?
1.computational statistics
2.computational artifacts(人工制品) that learn over time based on experience
一、分類
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
1.1 Supervised learning——Approximation
- 一句話實質：About Function Approximation(函數逼近)，or Approximate function induction(近似函數歸納)
- feed with labeled examples，comeing up with some function that generalizes beyond(泛化函數)
- 有反饋
1.2 Unsupervised learning——Description
- 一句話實質：About Compact(簡潔的) Description
- 無監督學習是密切相關的統計數據密度估計的問題。
- 無反饋
- Unsupervised learning could be helpful in the supervised Setting
1.3 Reinforcement learning (增強學習)
- 一句話實質：Learning from delayed reward (通過延遲性獎勵進行學習)
- 執行許多步之后才知道反饋，就像下棋（對比監督學習的立即反饋）
二、歸納法(induction)與演繹法(deduction)
- Generalize 泛化
- 了解機器學習發展史
- 機器學習算法與歸納而不是演繹有關
- Inductive bias 歸納偏差
歸納：從示例到一般規律（從一個示例得出更普遍的規律）
演繹：從規則到實例，a general rule to specific instances,basically like reasoning(推理)
三、三種機器學習的比較
表述成：優化問題
Supervised Learning —— labels data well(to find a funtion to score that) （標記數據）
Unsupervised Learning —— cluster scores well(最好的分類方法)
Reinforcement learning —— behavior scores well (最好的表現)
3.2 Data
Data is king in machine learning.
轉變：以算法為中心——》以數據為中心
- Believe in your data！