1. Review of Probabilities and Statics
1) Elements of Probability
- Random experiment : ์ํ, Nondeterministic
- Sample space : ์ ์ฒด์งํฉ ์์๋ค. Mutually exclusive(๋ฐฐ๋ฐ์ฌ๊ฑด)์ด๊ณ exhaustive(ํฉ์งํฉ = ์ ์ฒด์งํฉ) ์ด๋ค.
- Events : ์ฌ๊ฑด, Sample space์ ๋ถ๋ถ์งํฉ
- Probability : ํ๋ฅ . the proportion, or relative frequency (๋น์จ, ์๋์ ๋น๋์) / degree of belief (๊ธฐ๋๊ฐ)
2) Axioms of Probability Theory
- ํ๋ฅ ์ 0๊ณผ 1 ์ฌ์ด์
- Sample space์ ๋ชจ๋ ํ๋ฅ ์ ๋ํ๋ฉด 1
- P(a v b) = P(a) + P(b) - P(a^b)
3) Random Variable and Probability Density
- Random variable
* Domain : ์ ์์ญ
โ Discrete : ์ด์ฐ์ ํ๋ฅ ๋ณ์
โก Continuous : ์ฐ์์ ํ๋ฅ ๋ณ์
- Probability density function (pdf) : ํ๋ฅ ๋ฐ๋ํจ์
P(X=x) = P(x) = lim(dx->0) P(x<X<x+dx)/dx
- Probability mass function (pmf) : ํ๋ฅ ์ง๋ํจ์
P(X = 3) = P(3) = 1/6
- Cumulative probability density function(cdf) : ๋์ ๋ถํฌํจ์
Fx(x) = P(X<=x) = ∫P(u)du
4)
- Prior probability (๊ทธ๋ฅ ํ๋ฅ ํ๋ ๊ทธ ์์ฒด) : P(a)
- Posterior probability (conditional probability, ์กฐ๊ฑด๋ถ ํ๋ฅ ) : P(a|b) = P(a^b)/P(b)
- Joint probability (๋์์ ์ผ์ด๋ ํ๋ฅ ) : P(a^b) = P(X = a, Y = b)
- Product rule = P(a^b) = P(a|b) P(b)
- Marginal probability
P(X) = ∑ P(X , y) = ∑ P(X | y)P(y)
**************P(X ^ Y) = P( X, Y ) = P(X | Y) P(Y) ******************
5) Chain Rule
P(x1 , x2, x3, ... , xn) = P(xn | x1, ..., xn-1)P(x1, ..., xn-1) = P(xn| x1, ... , xn-1)P(xn-1 | n1, .., xn-2) P(x1 , ... , xn-2)
= P(xn | x1, ..., xn-1)P(xn-1|x1, ..., xn-2) ... P(x2|x1)P(x1)
= ∏P(xi | x1, ... , xi-1)
6) Independence
- Independence (๋ ๋ฆฝ) : P( a|b ) = P(a), P(a ^ b) = P(a)P(b)
- Conditional independence
: P(X, Y) != P(X)P(Y) (X์ Y๋ ๋ ๋ฆฝ์ด ์๋)
P(X,Y|Z) = P(X|Z)P(Y|Z) (Z๊ฐ ๋ผ๋ฉด X์ Y๊ฐ ์๋ก ๋ฌด๊ด, ์ฆ ๋ ๋ฆฝ์ด ๋จ)
P(X|Y) != P(X)
P(X| Y,Z) = P(X|Z)
ex) X: ํ์ด ์ํ
Y: ๋น๊ฐ ์ด
Z: ์ฐ์ฐ์ ์
Z๊ฐ ์์ ๋๋ Y(๋น๊ฐ์ด)->(์ฐ์ฐ์ ์)->X(ํ์ด ์ํ) ์ผ๋ก X์ Y๊ฐ Dependence ํ์ผ๋
Z๊ฐ ์ฃผ์ด์ง๋ฉด Y(๋น๊ฐ์ด)->Z(์ฐ์ฐ์ ์) // Z(์ฐ์ฐ์ ์) -> X(ํ์ด ์ํ) ์ผ๋ก X์ Y๊ฐ ๋ณ๊ฐ๊ฐ ๋จ
7) Expectation
- Expectation : E(X) = ∑ xi P(X=xi) = μx (ํ๋ฅ ๋ณ์์ ๊ฐ x ํ๋ฅ ์ ์ด ํฉ)
- Covariance (๊ณต๋ถ์ฐ) : cov(X, Y) = E((X - μx)(Y - μy)) : X๊ฐ ๋ณํจ์ ๋ฐ๋ผ Y๊ฐ ์ผ๋ง๋ ๋ณํ๋์ง expectation
* ๊ทธ๋ฅ ๋ถ์ฐ = E[(X - μx)^2 ]
- Covariance matrix ∑ (๊ณต๋ถ์ฐ ํ๋ ฌ)
: ∑ij = cov(Xi, Xj) = E( (Xi - μx)(Xj - μj ) )
๊ณต๋ถ์ฐ ํ๋ ฌ์ ( i, j ) ๊ฐ์ Xi ์ Xj์ ๊ณต๋ถ์ฐ
8) Gaussian Distribution ( ์ ๊ท ๋ถํฌ)
* Multivariate Gaussian distribution : ๋ค๋ณ๋ ์ ๊ท๋ถํฌ
* ๋ถ์ฐ -> vector (covariance matrix)
9) Central limit Theorem ( ์ค์ฌ ๊ทนํ ์ ๋ฆฌ)
: n๊ฐ์ ๋ ๋ฆฝ random variable์ ํ๊ท ์ n์ด ๋ฌดํ๋๋ก ๊ฐ๋ฉด gausian distribution์ ๋ฐ๋ฅธ๋ค.
2. Probabilistic Inference
: ๊ด์ธกํ evidence๊ฐ ์์ ๋, ์๊ณ ์ถ์ query proposition์ด true์ผ posterior probability๋ฅผ ๊ตฌํ๋ ๊ฒ
* Bayes' rule : ๋ฒ ์ด์ฆ ๊ท์น
- P(Y | X) = P(X|Y)P(Y) / P(X)
- P(Y | X, e) = P(X|Y,e)P(Y|e) / P(X|e)
Y : cause, ์์ธ, query proposition, ์๊ณ ์ถ์ ๊ฒ
X : ๊ฒฐ๊ณผ, ์ฆ์, evidence
์ ? P(Y|X) = P(X,Y)/P(X)
P(X|Y) = P(X, Y)/P(Y)
-> P(X,Y) = P(Y|X)P(X) = P(X|Y)P(Y)
-> P(Y| X) = P(X|Y)P(Y) / P(X)
๋๋ P(X, Y) = P(X | Y)P(Y)
์ฆ P(Y | X) = P(X, Y)/ P(X) = P(X|Y)P(Y) / P(X)
* Naive Bayes model
P(C, E1, E2, ... , En)
= P(C) P(E1, E2, ... , En|C)
= P(C) ∏P(Ei|C) (Ei ๋ผ๋ฆฌ๋ ์กฐ๊ฑด๋ถ๋ ๋ฆฝ์)
P(C | E1, E2, ... , En)
= P(E1, E2 , ... , En| C)P(C) / P(E1, E2, .. , En) <- Bayes rule
= αP(C) ∏P(Ei|C)
( α = 1 / P(E1, E2, .. , En) )
'๐ก๐ธ๐ธ๐ถ5: ๐ฆ๐๐๐๐ถ ๐ฐ๐๐พ๐ > Artificial Intelligence(COSE361)' ์นดํ ๊ณ ๋ฆฌ์ ๋ค๋ฅธ ๊ธ
[์ธ๊ณต์ง๋ฅ] 15. Probabilistic Reasoning over Time (PRoT) (0) | 2021.06.13 |
---|---|
[์ธ๊ณต์ง๋ฅ] 14. Bayesian Networks (0) | 2021.06.12 |
[์ธ๊ณต์ง๋ฅ] 8~9. First-Order Logic(FOL) (2) | 2021.04.25 |
[์ธ๊ณต์ง๋ฅ] 7. Propositional Logic - 3 (2) | 2021.04.25 |
[์ธ๊ณต์ง๋ฅ] 7. Propositional logic - 2 (1) | 2021.04.25 |