Definition of AI
Acting humanly: The Turing Test approach
Thinking humanly: The cognitive modeling approach
Thinking rationally: The “laws of thought” approach(Logic)
Acting rationally: The rational agent approach |
Turing test
natural language processing to enable it to communicate successfully in English;
knowledge representation to store what it knows or hears;
automated reasoning to use the stored information to answer questions and to draw new conclusions;
machine learning to adapt to new circumstances and to detect and extrapolate patterns.
total Turing test
computer vision to perceive objects;
robotics to manipulate objects and move about. |
Agents and Environments
An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through actuators. We use the term percept to refer to the agent’s perceptual inputs at any given instant. An agent’s percept sequence is the complete history of everything the agent has ever perceive. |
PEAS
Performance measure, Environment, Actuators, Sensors |
Knowledge Base
Conceptually, a set of sentences;
.Defined by TELL/ASK interface |
Declarative Approach
Program an agent by TELLing it things
Equivalently, construct and install a KB
Advantages
1.Flexibility: knowledge independent of how it would be used
2.Transparency to humans
3.Agent behavior malleable via language |
Knowledge Representation
Knowledge representation language: notation for expressing a KB
Consist of
Syntax: defines the legal sentences
Semantics: facts in the world to which sentences correspond, an interpretation for symbols in the logic
Logic: KR language with well-defined syntax and semantics
Model: Specifies truth or falsity of sentences |
Properties of Sentences
Valid: True in all models
Satisfiable: True in some model |
|
|
Rational agent
For each possible percept sequence, a rational agent should select an action that is expected to maximize its performance measure, given the evidence provided by the percept sequence and whatever built-in knowledge the agent has. |
Autonomy
A system is autonomous that its behavior is determined by its own percepts, rather than the prior knowledge of its designer. |
Coherence
Actions must be consistent with beliefs and desires. Necessary for knowledge-level understanding of computer programs. |
Agent Designs
Table Lookup, Simple Reflex, Model-based, Goal-based, Utility-based, Learning |
Environment Properties
fully/partially observable, single/multi, deterministic/stochastic, episodic/sequential, static/dynamic, discrete/continuous, known/unknown |
Admissibility and Consistency
Admissibility: h(n) ≤ c(n), where c(n) is the true cost of a solution along the current path through n.
Consistency: h(n) ≤ c(n,a,n')+h(n').
Optimality: If h is admissible and consistent, A* is optimal.
A* is optimal, complete and optimally efficient among all algorithms that extend search paths from root. |
Entailment and Inference
Entailment: α |= β if and only if M(α) ⊆ M(β). Truth of β is contained in α. α |= β if and only if the sentence (α ⇒ β) is valid. α ≡ β if and only if α |= β and β |= α.
Inference: α |- β means β can be derived from α.
An inference algorithm that derives only entailed sentences is called sound.
An inference algorithm is complete if it can derive any sentence that is entailed |
FOL to CNF
1.Translate bidirectionals to implications
2.Translate implications to disjunctions
3.Move negations inward (Use De Morgan’s laws until only atoms are negated)
4.Standardize variables
5.Eliminate existentials via skolemization
6.Drop universal quantifiers
7.Distribute and associate into CNF |
|
|
Uninformed search strategies
Breadth-first, Uniform-cost, Depth-first, Depth-limited, Iterative-deepening, Bidirectional, Graph search |
Analysis
Criterion |
BFS |
UC |
Complete |
Yes |
Yes |
Optimal |
Yes |
Yes |
Time |
O(bd+1) |
O(b1+floor(C*/e)) |
Space |
O(bd+1) |
O(b1+floor(C*/e)) |
Analysis
Criterion |
DFS |
DLDFS |
IDDFS |
Complete |
No |
No |
Yes |
Optimal |
No |
No |
Yes |
Time |
O(bm) |
O(bL) |
O(bd) |
Space |
O(bm) |
O(bL) |
O(bd) |
Informed Search
Priority-first(PFS), Greedy, A*, Iterative-deepening A*, heuristics and admissibility/consistency |
Flavors of PFS
Number of edges from origin: BFS
g(n), the path cost from the initial state(node) to node n: UCS(Dijkstra)
h(n), the estimated path cost from node n to goal: Greedy Search
f(n)=g(n)+h(n), the estimated cost of a path passing through n: A* search |
Memory-bounded heuristic search
iterative-deepening A* (IDA*)
Recursive best-first search (RBFS): best first search with linear space |
Local search strategies
Hill-climbing/stochastic, beam, simulated annealing, genetic algorithms |
Local search summary
Key advantages
1.Very little memory
2.Can often find reasonable solutions in large or infinite state spaces where systematic approaches are unsuitable
3.Better answers the more time spent
Disadvantages
Usually incomplete and not optimal |
Constraint propagation
Definition: Propagating the implications of a constraint on one variable onto other variables.
k-consistent: for any consistent assignment of k-1 variables, exists consistent value of any kth(Node/Arc/Path)
Strongly k-consistent: j-consistent for all j less than or equal to k |
Constrain Satisfaction Search
Backtracking
Minimum remaining values heuristic: choose variable with fewest legal values remaining
Degree heuristic: choose variable with largest number of constraints
Forward checking: delete inconsistent values ahead
Backjumping
Basic: backtrack to most recent decision
Conflict-Directed: backtrack to most recent variable in conflict with |
|
Created By
Metadata
Comments
No comments yet. Add yours below!
Add a Comment