AI Final
With regard to the stanford certainty factor algebra, be able to determain certainty factors when a rule fires (shown on page 351)
"CF(P1 ^ P2) = Min(P1, P2).
(∀ X) (clear(X) ←¬ (∃ Y) (on(Y,X)))
If block X is clear, there does not exist aby block Y such taht Y is on top of X
Know what makes a search algorithm admissible
If it is guaranteed to find a minimal path to a solution when such a path exists
What is hypothesis discrimination?
In which when more than one hypothesis survies the testing phase, we must determine addditional information can be gathered to continue search for the fault
Definition of a ATMS
It allows to maintain and reason with a number of simultaneous, possibly incompatible, current sets of assumption. Otherwise it is similar to JTMS. It also does not recognize the meaing of sentences
"
It currently is in the CH6 Guide
in a conceptual graph, it it just says name of concept: x
It is generic
Know the Turing Test and that it is not always the criteria for success of an AI application
It is not a criteria for success, do no forget you fool!
Know the four types of operations that can be performed on existing conceptual graphs to form new ones
Join, Restrict, Simplify, Copy
Know what the frame problem is with regard to planners
Keeping track of what changes in the world
Know how a production system implementing a blackboard architecture works. Know what the blackboard is. Know what knowledge sources are, what they do, and how they are related to each other
Knowledge Experts
Given two belief sets (m1 and m2) be able to calculate a third (m3) using Dempstrer's rule. Be able to do this when m1 and m2 have no empty intersections and also when they do. th etwo problems will be like the two cases on page 362
M1 & M2's truth table probablilties multiplied equals M3's probabilities
Know what the problem of local maxima is
No child is better than its parent
Does development of expert systems follow the traditional waterfall system?
No it does not
What is defeasibility?
Nonmonotonic Reasoning. That is, new information may invalidate previous results
What is a refraction conflict resolution strategy
Once a rule fires, can't run it again until a condition for it changes
Be familiar with the heuristics discussed in section 8.2.4 of your text, sppecifically the ordering of conditions/premises for rules
Ordering Premises in such a way that they may eliminatec ertain things as early as possible (On the quiz the answer was ABC)
Know and be able to apply Bayes theorem
Pr(E|F) = pr(E|F)pr(F) ---------------- pr(E|F)pr(F) + Pr(E|*F)pr(*f)
What does a row in a triangle macro represent
Prerequisites to a macro
What are scenes in a script
Presents a temporal aspect of the script
What us a frame axiom?
Rules to tell what predicates describing a state are not changed by rule applications and are thus carried over intact to help describe the new state of the world
what is g*(n)
Shortest path from start to n
Know an dbe able to recognize the general form of bayes theorem when a set of multiple hypotheses (more than wto) partition the set
Similar to ^
What are the criteria for problem solvers in an intelligent agent-based system
Situated, Autonomous, Flexible, and Sociable
Know what a goa driven search is and know when this typ eof search is best used
Take the goal that we want to solve. see what rules or legal moves could be used to generate this goal and determin what conditions must be true to use them. Theorem proving
What does durative mean in a teleo-reactive planner
Teleo-actions are executed as long as the acitons' preconditions are met and the assoicated goal has not yet been acheived
Know what a data-driven search is and know ehn this type of search is best used
The Problem solver begins with the given facts of the problem and a set of legal moves or rules for changing state. All or most of the data are given in the initial problem statement
What are the advantages of a Case Based Reasoner?
The ability to encode hisotrical knowledge directly. Allows shortcuts in reasoning. Allows a system to avoid past errors and exploit past successes. Extensive analysis of domain knowledge is not required. Appropriate indexing strategies add nsight and problem solving power.
For rules-based expert systems, know how rules are chosen and applied for a goal-driven system
The goal expression is initially placed in working memory. The system matches rule conclusions with the goal, selecting on erule and placing ts premise in the working memory
What are Props in a script
Things that support the content of the script
What are the characteristics of a STRIPS triangle table macro
USed to determine when that macro operator could be used in building a plan. Saving the maco operators and reusing them increase the efficiency in its searching plan
Given a transition matrix for the markove model, be able to determine a state duration (page 375, 376)
Use the matrix given to input the probabilities of what is expected. multiply together to get result
often simply called production. a production is a condition-action pair and defines a single chunk of problem solving knowledge. The condition part of the rule is a pattern that determines when that rule may be applied to a problem instance. the action part defines the associated problem-solving step
What is the set of production rules
Can a durative action be interrupted?
Yes
Know the characteristics of a recursive search of state space. is tail recursion used? is the OPEN data structure needed?
ask someone
Know the formula of expectation of events
ex(E) = r[reward] * p(E)[possibility of event] + c[cost] * (1-p(E)) [possibility of event not happening]
(∀ X) (pickup(X) → (gripping(X) ← (gripping( ) ∧ clear(X) ∧ ontable(X)))).
for all blocks X, pickup(X) means gripping(X) if the hand is empty and X is clear
in a conceptual graph, if it says concept:jared
it is specific
Know and be able to apply the following formula for conditional probablity
p(b|a) = (p(A ^ B)) ------------- p(a)
What does a column in a triangle amacro represent
results of a macro
What are arcs/edges in a semantic network
semantic relationships
The recognize-act cycle
the control structure for a production system is simple: working memory is initialize with the beginning problem description. the current state of the problem solving is maintained as a set of patterns in working memery. these patterns are matched against teh conditions of the production rules; this produces a subset of the production rules, called the conlict set
What are the results of a script
the facts that are true once the script has been terminated
What is conflict resolution
Choses a rule from the conflict set for firing. conflict resolution strategies may be simple, such as selecting the first rule whose condition matches the state of the world, or may involve complex rule selection heuristics. thi
For rules-based expert systems, know how rules are chosen and applied for a data-driven system
Compare the contents of working memory with the conditions of each rule base according to the order of rules in the rules base. If the data in working memory supports a rule's firing the result is placed in working memory and then control moves to the next rule.
What are nodes/vertices in a sementic network
Concepts
In a conceptual grraph, know how concepts are represented as opposed to concept relations
Concepts are represented as rectangles. Relationships are represented as ovals
What is working memory
Contains a description of the current state of the world in a reasoning process
What is depth_first_search
Depth_first = Much more efficien tfor many branches, because it does not need to keep everything in memory
What are Entry Conditions
Descriptors of the world must be true for the script to be called
what is h(n)
Distance from state n to goal
Know what tradition of philosophy associationist theories follow
Empiricist Tradition
Know what a hill-climbing search algorithm is as well as the characteristics of it
Expand the current state of the search and evaluate children
Know the Three Primary people involved in building an expert system
Expert, Knowledge Engineer, and User
What is the recency resolution strategy
Favor rules whose conditions match the facts most recently added
What is the specificity resolution strategy
Favor the more specific rules (more conditions)
(∀ X) (∀ Y) (stack(X,Y) → ((on(X,Y) ∧ gripping( ) ∧ clear(X)) ← (clear(Y) ∧ gripping(X)))).
For all X and Y, if you stack X on Y, then X will be on Y, the hand will be empty, and there will be nothing on X, so long as Y is clear and hand is holding X
(∀ X)(∀ Y) (unstack(X,Y) → ((clear(Y) ∧ gripping(X)) ← (on(X,Y) ∧ clear(X) ∧ gripping( ))).
For all X and Y, if you unstack X from Y, then Y will be clear, and hand will be holding X, so long as X is on Y, X is clear, and hand is empty
(∀ X) (∀ Y) (∀ Z) (unstack(Y,Z) → (ontable(X) ← ontable(X))).
For all X, Y, and Z, if you onstack Y from Z, X will be on the table, so long as X was already on the table
(∀ X) (∀ Y) (∀ Z) (stack(Y,Z) → (ontable(X) ← ontable(X))).
For all X, Y, and Z, if you stack Y on Z, X will be on the table, so long as X is already on the table
(∀ X) (putdown(X) → ((gripping( ) ∧ ontable(X) ∧ clear(X)) ← gripping(X))).
For all X, if you putdown (x) , then hand will be empty, x will be on the table, and nothing will be on top of X, so long as X is being gripped
(∀ Y) (∀ X) ¬ (on(Y,X) ← ontable(Y))
For every Y an dX, there does not exist a Y on top of X that is also on the table
(∀ Y) gripping( ) ↔¬ (gripping(Y))
For every Y, gripping is empty so long as gripping does not have Y
What is hypothesis testing?
Given a collection of potential faulty components, we determing which of them would have explained the observed behaivour
What is hypothesis generation
Given a descrepency, we hypothesized which components of the device could have caused it
Know whether admissible algorithms are monotone or not
All monotone algorithms are admissible, not all admissible algorithms are monotone
Know what a heuristic is and why it is used
Any approach to problem solving that employs a practical method, not guaranteed to be optinmal, but instead sufficient for reaching an immediate goal
Definition of a BBN
Basic Chain Probability, uses table of probabilities
What is breadth first search
Breadth_first = shortest path
What the the disadvantages of a Case BAsed Reasoner?
Cases do not often include deeper knowledge of the domain. A large case base can suffer problems from store/compute trade-offs. It is difficult to determine good criteria for indexing and matching cases
What types of information is contained in a frame
1. Frame identification information 2. Relationship of this frame to other frames. 3. Descriptors of requirements. 4. Procedural information
Definition of a JTMS
A simple TMS where one can examine the consequences of the current set of assumptions. The meaning of sentences are not known
Know what an opportunistic search is
A simple search strategy, whenever a rule fires to conclude new informationm control moves to consider those rules which have that new information as a premise. This makes any new concluded information (search does not change as the result of "askable" premises) the controlling force for finding the next rules to fire. This is termed opportunistic because each conclusion of new information drives the search
What is the conflict set
A subset of the production rules. Whose conditions match the patterns in working memory. the productions in the conflict set are said to be enabled. One of the productionsin the conflict set is then selected (conflict resolution) and the production is fired
What is the general strategy of a teleo-reactive planner
Accomplish complex problem solving through the coordination of simple tast-specific agents. The justification for this approach is that simple agents have the advantage of working in smaller and more constrained problem spaces. The higher level controller can make more global decisions
What are Roles in a script
Actions that the individual participants perform
what is h*(n)
Actual cost of shortest path from n to goal
what is g(n)
Actual length of the path from any state n to the start state
