Accounting Uncertainty Chapter 12

Flashcards related to the Accounting Uncertainty Chapter 12 lecture notes.

What causes uncertainties in real-world problems?

Problems contain uncertainties due to partial observability, nondeterminism, or unexpected situations.

How can an agent account for the uncertainty of its own representation of the world?

Modify the agent function so that it selects rational actions based on its own state of internal belief in what is rational.

What is Probability theory?

deals with the degree of belief of relevant sentences rather than the firm knowledge of their truth and summarizes the uncertainty that comes from our ignorance

What is Utility Theory?

accounts preferences based on the quality of the outcome being useful where every state has a degree of usefulness for achieving the end goal (utility), and higher utility is preferred

When is an agent considered rational in Decision Theory?

If and only if it chooses the action that yields the expected utility and is also optimal of the choice is based on the maximum expected utility, MEU

How does probability theory consider the world?

A set of possible worlds Ω which are mutually exclusive

What is unconditional or prior probability?

The degree of belief that the propositions are true in the absence of any other information

What is conditional or posterior probability?

Given evidence that an event causes the observation a happening, it is the degree of belief that a new event will cause the event b to happen as well

What is Factored representation?

The possible world is represented by a set of variable/value pairs that characterize it, also known as random variables, and their names begin with an uppercase letter

What is Probabilistic Inference?

For any potential effect φ, we sum the events when the proposition φ is true to determine the unconditional probability for it to happen

What does the knowledge base represent?

Probability distribution of boolean variables representing the facts about the world

What does the decision base represent?

Probabilistic Inference for calculating the conditional probability distribution of the effects of the actions based on the joint unconditional distributions of their causes

Describe the problem-solving strategy of Inference Using Full Joint Distributions

Compute the distribution of all variables of interest by fixing the distribution of evidence variables and summing over the distribution of hidden variables

What is the application area of Inference Using Full Joint Distributions?

Medical diagnostics, unauthorized intrusion detection, online fraud detection, error correction, device malfunctioning, etc.

What is the Worst-case time complexity for Inference Using Full Joint Distributions?

O(dn) where d is the largest arity amongst the variables

When are a and b considered independent?

If P (a|b) = P (a) or P (b|a) = P (b) or P (a b) = P (a) P(b)

What is Bayes' Rule?

P (b | a) = P (a | b) P(b) / P (a)

What do conditional probabilities P (effect|cause) and P (cause|effect) quantify?

The conditional probability P (effect|cause) quantifies the relationship in the causal direction, whereas P (cause|effect) describes the diagnostic direction – applicable in both forward and backward search problem solving

What are Bayesian networks?

Represent dependencies amongst the variables with a simple, directed graph in which each node is annotated with quantitative probability information

What are Simple queries in Bayesian Networks?

Compute the posterior probability for facts Xi given all possible observations E

What are Conjunctive queries in Bayesian Networks?

Compute the probability for a joint cause given the posterior probabilities for each individual causes

What is the basic idea behind Approximate Inference for Bayesian Networks?

Draw N samples from a sampling distribution S and compute an approximate posterior probability P ˆ

What is a Markov chain?

A stochastic process describing a sequence of possible events

What is Markov chain Monte Carlo (MCMC)?

Is a sample from a stochastic process whose distribution is the true posterior probability of the nodes and is Used for simulating sampling from complex probability distributions due to its independence of the previous probabilities in the simulation loop

What do probabilities express?

Expresses the agent’s inability to reach a definite decision regarding the truth of a sentence.

What does decision theory combine?

Combines the agent’s beliefs and desires, defining the best action as the one that maximizes expected utility.

What does the full joint probability distribution specify?

Specifies the probability of each complete assignment of values to random variables