Module 5 Flashcards
What are knowledge based agents
Use process of reasoning to decide actions
What is a knowledge base
Set of sentences in formal language
What’s is the declarative approach to build KB agent
Tell what it needs to know
Ask itself what to do from KB
Operations involve inference
Can single inference algorithm answer any solvable question
Yes
What is inference engine and KB
KB - domain specific facts
Inference engine - generic code
Describe the KB agent flow
Given percept, agent adds percept to KB and asks KB for best action, it communicates KB that it took action
List types of KB agents
Declarative - build KB by telling it what it needs to know
Procedural - Encode desired behaviours as program code
How do agents acquire knowledge
Through percept - (transition model and sensor model)
Can keep track of partially observable world
What is syntax
What sentences are allowed, defines well formed sentences
What is semantics
Meaning of sentence linked to the world, where it’s true / false
What is entailment
One thing logically follows another
When does KB entail a
Iff a is true in all worlds where KB is true
What is a model
M is a model of a sentence a if a is true in m
What is M (a)
The set of all models of a
What is inference
Allows new sentences to be derived from KB
Inference notation
KB |-I a
What does KB |-I a mean
Sentence a can be derived from KB by using inference I
What is the soundness of inference
Everything probable is true, no false positives, derived entailed sentences. KB |-i a true then KB |= a is true
What is the completeness of inference
Everything true is possible, no false negatives. KB |= a true then KB |-I a is true
KB transference into real world
If KB is true in real world, then any sentence a derived from KB by a sound inference is also true in real world
Types of inference proofs
Model checking - for every possible world, if a is true make sure b true aswell
Theorem proving - search for a sequence of proof steps leading from a to b
List the types of logic
Propositional
First order logic
Relational db
What is propositions calculus and what does proposition help with
Proposition helps with inference and and propositional calculus is language of propzrisitons
What is a well formed formula
An atom.
What does PL deal with
Validate and unsatisfiability of a formula and derivation of new formulas
What is a tautology
Iff a is true in all interpretations
When is a formula valid
Valid iff tautology
When is formula consistent
If at least one interpretation true
When is formula inconsistent/unsatisfiable?
Iff a is always false in all interpretations
Truth table pros and cons
Pros - good for small number of clauses and good at representing survey
Cons - n atoms lead to 2^n entries, all entries may not be relevant
Reasoning pattern for inference
And elimination
Modus pones
De Morgan’s law
Monotonic, sound but not complete
Resolution refutation
Negation of goal is proved by showing empty set.
Conjunctive normal form
( V V V ) /\ ( V V V)
Conversion to CNF
eliminate -> and
Eliminate ~~
Use de Morgan
Use distributive
If two clauses contains complementary literals
Can be resolved
New clauses is resolvent to parent clauses
Another name for resolvent
Logical consequence
