EECS492Mid Cheat Sheet by kirawen

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

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.

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.

Perfor­mance measure, Enviro­nment, Actuators, Sensors

Concep­tually, a set of sentences;
.Defined by TELL/ASK interface

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 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

Valid: True in all models
Satisf­iable: True in some model

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.

A system is autonomous that its behavior is determined by its own percepts, rather than the prior knowledge of its designer.

Actions must be consistent with beliefs and desires. Necessary for knowle­dge­-level unders­tanding of computer programs.

Table Lookup, Simple Reflex, Model-­based, Goal-b­ased, Utilit­y-b­ased, Learning

fully/­par­tially observ­able, single­/multi, determ­ini­sti­c/s­toc­hastic, episod­ic/­seq­uen­tial, static­/dy­namic, discre­te/­con­tin­uous, known/­unknown

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: 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

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

Breadt­h-f­irst, Unifor­m-cost, Depth-­first, Depth-­lim­ited, Iterat­ive­-de­epe­ning, Bidire­cti­onal, Graph search

Priori­ty-­fir­st(­PFS), Greedy, A*, Iterat­ive­-de­epening A*, heuristics and admiss­ibi­lit­y/c­ons­istency

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

iterat­ive­-de­epening A* (IDA*)
Recursive best-first search (RBFS): best first search with linear space

Hill-c­lim­bin­g/s­toc­hastic, beam, simulated annealing, genetic algorithms

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

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

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