- 朴素贝叶斯法的三要素:学习输入/输出的联合概率分布、后验概率最大化、通过极大似然估计或贝叶斯估计进行参数求解
一、储备知识
- 全概率公式:完备事件组 A 1 , ⋯ , A n A_1,\cdots,A_n A1,⋯,An, A 1 ∪ ⋯ ∪ A n A_1\cup\cdots\cup A_n A1∪⋯∪An, A i A j = ∅ , i ≠ j A_iA_j=\varnothing,i\neq j AiAj=∅,i=j,有
B ∩ Ω = B ∩ ( A 1 ∪ ⋯ ∪ A n ) = ( B A 1 ) ∪ ⋯ ∪ ( B A n ) \begin{aligned} B\cap \Omega&=B\cap(A_1\cup\cdots\cup A_n) \\ &=(BA_1)\cup\cdots\cup (BA_n) \\ \end{aligned} B∩Ω=B∩(A1∪⋯∪An)=(BA1)∪⋯∪(BAn)且 ( B A i ) ∩ ( B A j ) = ∅ , i ≠ j (BA_i)\cap(BA_j)=\varnothing,i\neq j (BAi)∩(BAj)=∅,i=j,从而
P ( B ) = P ( A 1 B ) + ⋯ + P ( A n B ) = ∑ i = 1 n P ( A i B ) = ∑ i = 1 n P ( A i ) P ( B ∣ A i ) \begin{aligned} P(B)&=P(A_1B)+\cdots+P(A_nB) \\ &=\sum_{i=1}^nP(A_iB) \\ &=\sum_{i=1}^nP(A_i)P(B|A_i) \\ \end{aligned} P(B)=P(A1B)+⋯+P(AnB)=i=1∑nP(AiB)=i=1∑nP(Ai)P(B∣Ai) - 贝叶斯定理: P ( A i ∣ B ) = P ( A i B ) P ( B ) = P ( A i ) P ( B ∣ A i ) P ( B ) P(A_i|B)=\frac{P(A_iB)}{P(B)}=\frac{P(A_i)P(B|A_i)}{P(B)} P(Ai∣B)=P(B)P(AiB)=P(B)P(