Date: 23.04.25 ~ 23.04.31
Writer: 9tailwolf
Introduction
In this chapter, we learn about probabilistic reasoning
. I introduced about the way that can representate relationship, and some algorithms.
Followings are some probabilistic inference tasks.
- Simple queries
- Conjunctive queries
- Optimal decisions
- Value of information
- Sensitivity analysis
- Explanation
Bayesian Networks
Bayesian Networks
is a graph that representate events’ relationship. There is some characteristic.
- A direct grpah.
- An acyclic graph.
- A conditional distribution for each node given its parents : \(P(X_{i}\mid Parent(X_{i}))\).
- The calculation is a sum of all products with the number of cases and probability of events.
CPT : Conditional probability table
Global Semantics
\(P (x_{1},x_{2},...,x_{n}) = \Pi_{i=1}^{n} P(X_{i}\mid Parent(X_{i}))\)
Local Semantics
\(P (x_{1},x_{2},...,x_{n}) = P(X_{n}\mid Parent(X_{n-1}))\)
The idea of local semantics came from Markov Chain
, which suppose that past can’t affect event of future.
Enumeration Algorithm
Algorithm
Variable Elimination Algorithm
Algorithm
Inference by Stochastic Simulation
Prior Sampling
Reject Sampling
Likehood Weighting
Gibbs Sampling