ARTIFICIAL INTELLIGENCE
UNIT-1
1. WHY
GO IN FOR AI ?
To learn more about ourselves and to build intelligent
entities that can perceive
objects,
remember them , take decisions based on past occurrences and reason
out
things in the manner we do.
2. WHAT
IS AI?
a. Art of creating machines that perform functions that
require intelligence
when performed by people.
b. Study of computations that make it possible to perceive,
reason,and act.
3. WHAT
WAS TURING TEST AIMED AT?
It was aimed at deciding if a computer was capable to
achieve human level
performance
in all cognitive tasks. If so, then the computer was termed intelligent.
4. HOW
WAS TURING TEST DESIGNED?
The computer to be tested is interrogated by a human via a
teletype and the
computer
passes the test if the interrogator cannot tell if there is a human or
computer
at the other end.
5.
REQUIRED CAPABILITIES OF COMPUTER TO PASS TURING TEST?
a. Natural language processing.
b. Knowledge representation
c. Automated reasoning
d. Machine learning
6.
NATURAL LANGUAGE PROCESSING :
To enable the machine to communicate successfully in
English or some human
language.
7.
KNOWLEDGE REPRESENTATION:
To store information provided before or during the
interrogation i.e., information
gained by past experiences.
8.
AUTOMATED REASONING:
To use the knowledge base as an inference material, to
answer questions and to
draw
new conclusions.
9.
MACHINE LEARNING:
To adapt to new circumstances and to detect and extrapolate
patterns.
10.
OBJECTIVE OF TOTAL TURING TEST:
In this test including the above listed capabilities the
interrogator can also test the
perceptual
abilities and recognitional capabilities.
To pass
the test the computer will need
a. computer vision: to perceive objects.
b. Robotics : to move them about.
11.
STATE THE MAIN PURPOSE OF COGNITIVE SCIENCE:
It aims at constructing precise theories of the workings of
the human brain and
later
express these theories as a computer program.
12.
WHAT IS AN AGENT?
An agent is an entity that can perceive its environment
through sensors and acting
upon
that environment through effectors.
13. WHY
IS AN AGENT CALLED RATIONAL?
An agent perceives and acts. Acting rationally is to
achieve one’s goals, given
one’s
beliefs. Hence the agent is called rational when it reasons logically to
conclude
that a given action will achieve one’s goal and then act on that
conclusion.
14. HOW
AND WHEN TO EVALUATE AN AGENT’S SUCCESS?
Performance measure will decide how successful an agent is.
The
period of time during which the perfomance of the agent is evaluated is the
when.
15.
DEFINE IDEAL RATIONAL AGENT
Depending on the built in knowledge the agent has an ideal
rational agent should
do
whatever action is expected to maximize its performance measure.
16.
WHEN IS A SYSTEM CALLED AUTONOMOUS?
A system is autonomous when its behaviour is determined by
its own experience.
17.
WHAT IS AN AGENT PROGRAM?
It is a function that implements the agent mapping from
percepts to actions. The
program
is run on a computing device other wise called an architecture.
18.
LIST DIFFERENT TYPES OF AGENT PROGRAM.
. simple reflex agents
. agents that keep track of the world
. goal-based agents
. utility based agents
19.
NATURE OF WORK DONE BY SIMPLE REFLEX AGENT
It works by finding a rule whose condition matches the
current situation and
then
doing the action associated with that rule.
20.NATURE
OF WORK DONE BY AGENTS THAT KEEP TRACK OF THE
WORLD.
It works by finding a rule whose condition matches the
current situation (as
defined
by the percept and the stored internal state )and then doing the action
associated
with
that rule.
21.WORK
DONE BY GOAL BASED AGENTS.
In addition to the current state description the agent
requires goal information
which
describes situations that are desirable. Based on this the agent can do
actions
that result in achieving the prescribed goal.
22.
WORK DONE BY UTILITY BASED AGENTS.
Utility is a function that maps a state onto a real number,
that describes
associated
degree of happiness. An agent that posseses an explicit utility function can
make
rational decision.
23. HOW
DOES THE ENVIRONMENT THE AGENT IS IN AFFECT ITS
WORKING?
Actions are done by the agent on the environment ,which
provides percepts to the
agent.
24.
PROPERTIES OF ENVIRONMENT
a. accessible or inaccessible
b. deterministic or nondeterministic
c. episodic or nonepisodic
d. static or dynamic
e. discrete or continuous
25.
ACCESSIBLE OR INACCESSIBLE ENVIRONMENT
When the agents sensory apparatus gives it access to the
complete state of
environment
then the environment is termed accessible, else it is termed inaccessible.
26.
DETERMINISTIC OR NON DETERMINISTIC
If the next state of environment is completely determined
by the current state and
the
actions selected by the agents, then we say environment is deterministic,else
non
deterministic
27.EPISODIC
OR NON EPISODIC
Each episode consists of agent perceiving and then acting.
Quality of action
depends
on the episode itself and not on the previous episodes.
28.
STATIC OR DYNAMIC
If environment can change while an agent is deliberating
then we say the
environment
is dynamic for the agent else static.
29.
DISCRETE AND CONTINUOUS
If there a limited number of fixed and distinct percepts
and actions the
environment
is discrete ,else continuous.
30.
PROBLEM SOLVING AGENTS
These agents decide what to do by finding sequences of
actions that lead to
desirable
states.
31.
STEPS IN PROBLEM SOLVING
a. goal formulation
b. problem formulation
32. WHY
SEARCH ALGORITHM’S ARE USED?
An agent with several immediate options of unknown value
can decide what to
do by
first examining different possible sequences of actions that lead to state of
known
value, and then choosing the best one. This process requires application of
search
algorithm.
33.
WHAT ARE THE DIFFERENT PROBLEM TYPES.
a. single-state problem
b. multiple state problem
c. contingency problem
d. exploration problem
34. WHY
ARE SEARCH STRATEGIES USED?
When a search is conducted the choice of which state to
expand first is
determined
by the search strategy.
35. HOW
ARE SEARCH STRATEGIES EVALUATED?
a. completeness
b. time complexity
c. space complexity
d. optimality
36.
LIST DIFFERENT SEARCH STRATEGIES.
a. breadth- first search
b. uniform cost search
c. depth first search
d. iterative deepening search
e. bi-directional search
37.
BREADTH FIRST SEARCH
All the nodes at depth d in the search tree are expanded
before the nodes at depth
d+1.
38.
UNIFORM COST SEARCH
It modifies the breadth first strategy by always expanding
the lowest cost node on
the
fringe rather than the lowest depth node.
39.
DEPTH FIRST SEARCH
This search always expands one of the nodes at the deepest
level of the tree,when
the
search hits a dead end the search goes back and expands nodes at the shallow
level.
40.
ITERATIVE DEEPENING SEARCH.
It combines the benefits of depth first and breadth first
search .the order of
expansion
of states is similar to breadth first except that some states are expanded
multiple
times.
41.
BIDIRECTIONAL SEARCH
In this the search is conducted simultaneously both forward
from the initial state
and
backward from the goal, and stop when the two searches meet in the middle.
42.
GREEDY SEARCH
One of the simplest best first search strategies is to
minimize the estimated cost
to
reach the goal,that is the node whose state is judged to be the closest to the
goal
state is always expanded first.
43. WHY
OPT GAMES AS SEARCH PROBLEMS?
Games have engaged the intellectual faculties of human to
an alarming degree as
long as
civilization has existed. Abstraction of the game make it appealing for
conducting
research.
UNIT – 2
1. What
is the aim of knowledge based agents?
It aims to implement a view of agents in which they can be
seen as knowing about
their
world, and reasoning about their possible courses of action.
2. What
are the two elements of logic?
_ a formal language in which knowledge can be expressed
_ a means of carrying out reasoning in such a language
3. What
is the central component of a knowledge-based agent?
_ knowledge base
4. What
is knowledge base or KB?
A knowledge base is a set of representations of facts about
the world.Each individual
representation
is called sentence.
5.What are the three levels of knowledge based agent?
• Knowledge level or Epistemological level
• Logical level
• Implementation level
6.What
is the object of knowledge representation?
The object of knowledge representation is to express
knowledge in computer-tractable
form,
such that it can be used to help agents perform well.
7. What
are the two aspects of knowledge representation language?
• Syntax
• Semantics
8. What
is entailment?
We want to generate new sentnces that are necessarily true,
given that the old sentences
are
true.This relation between sentences is called entailment.
9. What
is sound or truth-preserving?
An inference procedure that generated only entailed
sentences is called sound or truthpreserving.
10.
What is proof?
The record of operation of a sound inference procedure is
called a proof.
11.
When is an inference procedure complete?
An inference procedure is complete if it can find a proof
for any sentences that is
entailed.
12.
What should a good knowledge representation do?
_ should combine the advantages of natural languages and
formal languages.
_ should be expressive and concise so tat we can say
everything we need to say
succinctly.
_ should be unambiguous and independent of context.
_ should be effective.
13.
What is compositional language?
An compositional language is one in which the meaning of a
sentence is a function of the
meaning
of its parts.
14.
When is a sentence true?
A sentence is true under a particular interpretation if the
state of affairs it represents is the
case.
15.
When is a sentence valid?
A sentence is valid or necessarily true if and only if it
is true under all possible
interpretations
in all possible worlds.
16.
When is a sentence satisfiable?
A sentence is satisfiable if and only if there is some
interpretation in some world for
which
it is true.
17.
What does a logic consist of?
(1) (a) syntax of the language
(b) semantics of the language
(2) Proof theory
18.
What is model?
Any world which a sentence is true under a particular
interpretation is called a model of
that
sentence under that interpretation.
19.
What are the seven inference rules for propositional logic?
• Modus Ponens or Implication-Elimination
• And-Elimination
• And-Introduction
• Or-Introduction
• Double-Negation Elimination
• Unit Resolution
• Resolution
20.
What is first order logic?
First Order Logic is a general purpose representation
language that is based on an ontological
commitment
to the existence of objects and relations in the world.
21. What
is an atomic sentence?
An atomic sentence consists of a predicate applied to one
or more terms; it is true just
when
the relation named by the predicate holds between the objects named by the
terms.
22.
What are the two standard quantifiers of first order logic?
• Universal
• Existential
23.
What is ground term?
A term with no variables is called a ground term.
24.
What are diachronic rules?
Rules describing the way in which the world changes ( or
does not change ) are called
diachronic
rules.
25.
What is situation calculus?
Situation Calculus is the name for a particular way of
describing change in first-order
logic.
26.
What are synchronic rules?
The axioms written to capture the necessary information for
these deductions are called
synchronic rules, because they relate properties of a world
to other properties of the sameworld state.
27.
What are the two main kinds of synchronic rules?
• Causal rules
• Diagnostic rules
28.
What is knowledge engineering?
The process of building a knowledge base is called
knowledge engineering.
29.
What is partition?
A
disjoint exhaustive decomposition is known as partition.
30.
What are the three new inference rules?
• Universal Elimination
• Existential Elimination
• Existential Introduction
31.
What is forward chaining?
We can start with the sentences in the knowledge base and
generate new conclusions that in turn allow more inferences to be made. This is
called forward chaining.
32.
What is backward chaining?
We can start with something we want to prove, find implications
sentences that would allow us to conclude it, and then attempt to establish
their premises in turn. This is called backward chaining.
33.
What is Conjunctive normal form(CNF)?
All the disjunctions in the KB are assumed to be joined in
one big, implicit conjunction , so this worm is called conjunctive normal
form(CNF).
34.
What is refutation?
One complete inference procedure using resolution is
refutation, also known as proof by
contradiction.
UNIT 3
*** RULES AND RULE CHAINING***
1. What
is meant by rule based system?
Rule based problem solving systems are those which are
built using rules
like
if-then patterns.
2. What
is meant by the following terms?
Assertion : Statement which is true.
Working memory: Collection of assertions.
Antecedent: the convention used to represent if- patterns.
Consequent: “ “ “ “ then- patterns.
3. What
is a deduction system?
It is a rule based system in which then-patterns specify
new assertions to
be
placed in working memory.
4. What
is a reaction system?
Here the then-patterns specify actions rather than
assertions.
5. What
is forward chaining?
It is the process of moving from the if patterns to the
then patterns using
the if
patterns to identify appropriate situations for the deduction of a new
assertion.
6. When
a rule is triggered,fired or intiated?
When an if pattern matches an assertion,the antesedent is
satisfied.
When all if patterns of a rule are satisfied, the rule is
triggered.
If a triggered rule establishes a new assertion or action,
it is fired.
7. What
is backward chaining?
It is the process of using antecedent-consequent rules to
work backward
toward
hypothesis supporting assertions.
8. Give
some application of rule based deduction system.
MYCIN and XCON.Mycin helps the physician to prescribe
disease
specific
drugs.Xcon works in a computer system component management domain.
9. What
is meant by conflict resolution procedure?
It is used to determine which rule to fire among the
triggered ones.One of
the
simplest procedure is “ Rule ordering” ,ie, arranging rules in a list and
the
first triggeres rule is the first one to be fired.
10.
List the various conflict resolution procedure.
Rule ordering, Context limiting,specifying ordering,
Data,Size,Recency
ordering.
11.
What is meant by rete procedure?
RETE(net) procedure names a proc. that works by moving each
newassertion,viewed
as a relational record,thro’ a rete of boxes,each
performing
some relational operation.
*** Rules Substrates and Cognitive Modeling ***
12.
What is meant by inference net?
Inference net can be used to produce a goal tree.Each node
in this tree
corresponds
to the application of an antecedent – cosequent rule.
13.
What r the two modes of running rule based sysems?
Show-me mode: Treat the providing assumptions as tho’ they
r ordinary
antecedents.
Ask-question-later
mode: Ignore all providing assumptions.
14.
Name the other different modes.
Decision maker mode,trusting skeptic mode,progressive
reliability mode.
15. Who
is a domain expert?
A domain expert is a person who has accumulated a great
deal of skill in
handling
problems in a particular area called “ Domain of Expertise”.
16.
What is meant by knowledge engineering?
It is the extraction of useful knowledge from domain
experts for use by
computers.
17.
What r the merits of rule based system?
• They solve many problems.
• They answer simple questions abt how they reach their
conclusions.
18.
What r the de-merits of rule based system?
• They do not reason on multiple levels.
• use constraint exposing models.
. look at problems
form different perspectives.
• know how n when to break thire own rules.
• have access to reasoning behind their rules.
19.
What is meant by Production systems,short term n long term memory?
In human modeling world,if-then rules r called productions
and rule based
systems
are called Production systems.
20.
What is meant by short term n long term memory?
Short term memory: It is the human thinking which involves
productions
triggered
by short living items. Long term memory: It holds all productions.
21.
What r the 2 protocols used by information processing psychologists?
The state of knowledge: what the subject knows The problem
–behaviour graph: trace of subject moving thro’ states of
knowledge.
22.
What is meant by architecture in AI ? Give an example
It is an integrated collection of representation and
methods to handle a
specified
class of problems.
Eg:
SOAR( State,Operator And Result)
23.
What is meant by preference net?
It is a net in which preference labels and preference links
describe the
merits
of various choices.
*** Frames and Common sense ***
24.
What is meant by Thematic Role frame?
It is an action oriented representation focussed on
identifying the roles
played
by various objects.
25.
List some thematic roles often used.
Agent,Coagent,Beneficiary,Thematic object,instrument,source
n
destination,
old and new surroundings,conveyance,trajectory,
location,time,duration.
26.
What r the 3 types of world used in thematic role frames?
Physical world : Objects change postions,acquire and lose
various properties.
Mental
world: Objects are facts,ideas and concepts.
Ownership
world: Objects are abstract certificates of control,ownership etc.,
27.
What is the use of primitive action frames and state change frames?
They are used to answer questions about what probably was
one or what
probably
happened next.
UNIT- 3
Semantic nets and
description matching
1. Define representation and description.
Ans: Representation is a set of conventions about how to
describe a class of
things.
Description makes use of the conventions to describe some particular
thing.
2. When
a problem is said to be solved?
Ans: Once a problem is described using an appropriate
representation, the
problem
is almost solved.
3. A
representation has 4 fundamental parts. What are they?
Ans: Lexical part-determines vocabulary, structural
part-describes constraints,
procedural
part-specifies access procedures and semantic part-associate meaning
with
description.
4. A
semantic net is a ________.
Ans:
representation
5. What
are the 3 semantics?
Ans:
Equivalence, procedural & descriptive semantics.
6.
_______________ is used to identify an object by first describing it and then
searching
for a match in a description library.
Ans:
Describe-and-match method
7. Give
one of the simplest application of describe-and-match method.
Ans:
Feature-based object identification. It consists of feature extractor & a
feature
evaluator. Values obtained by the feature extractor become the coordinates
of
feature point in feature space.
8. In
geometric analogy net two rule parts (rule descriptions are there)-what are
they?
Ans:
Object-relation description such as above, inside, to the left-of etc., and
object-transformation
description such as rotation, reflection, deletion, expansion
etc.
9.
Describe geometric analogy net.
Ans: It
is a representation & it is a semantic net in which the nodes are the
geometric
objects and edges describe their relation or transformation.
10. If
A is =B then c = 1 2
3 let
be l, and be m, be x & be y
Ans: c
= 3 (as l is above m in A, & its size are unchanged & transformed to B,
where l
is in left-of m. similar transformation is taken place between x&y in
c&3).
11. If
A is =B then c = 1 2
3 let
be rotated 45 degrees
Ans: c
= 1 as it is only rotated 45 degrees ( in 2 it is shrunk & in 3 it is left
unchanged).
12. If
A is =B then c = 1 2
3
Ans: c
= 1 as A’s inner square is deleted so as in 1
13. How
describe-and-match method is used to recognize abstractions in story plots?
Ans: We
need vocabulary of node types and link labels. There are 3 node types:
mental
state MS; positive event +; &negative event -. There are 3 link labels: i,
an
acronym
for initiates, (meaning that the MS or event at the tail of an i link leads to
the one
at the head of the link); t, for terminates, (meaning that the MS or event at
the
tail turns off the one at the head); & c, for co refers, (meaning that the
MS or
event
at the tail refers to the same MS or event as the one at the head), c is
doubleheaded.
There
are about 3*3*3 = 27 node-link-node combination of this 15 are
base
units & others are composite units & abstraction units.
14.
Explain about base units.
Ans:
Out of 27 node-link-node combinations 15 have a neutral, easily stated
interpretation,
and each of these is called a base unit. They are 3 groups one
involves
MS initiating event or vice versa (success, failure, enablement,
motivation)
another involves 2 MS (recursion, change of mind, perseverance)
&
finally group which does involve MS only events (positive tradeoff, loss,
resolution,
negative tradeoff, positive co reference, mixed blessing, hidden
blessing,
and negative co reference.
15.
Explain about complex units & abstraction units.
Ans:
Base units often overlap, producing recognizable aggregates & these
aggregates
are called composite units (fleeting success, success born of adversity,
fortuitous
success, intentional problem resolution).Together base units and
composite
units constitute abstraction units (retaliation).
Frames
and inheritance
1. Give
some of the problem-solving method.
Ans:
Describe & match method, inheritance, frames etc.,
2. CLOS
(Common Lisp Object System) inheritance procedure determines a
___________
among multiple classes.
Ans:
precedence ordering
3. A
semantic net can be viewed as a collection of __________.
Ans:
nodes and links/ frames
4. What
is a frame?
Ans:
Each node and its links can be collected together and called a frame. It uses
rectangle-and-slot
notation. The frame name is the name of the node & name
attached
to the slot is the names of the links emanating from a node. Slot values
represent
the destination of links emanating from a node.
5.
Define frame system.
Ans: It
is a semantic net in which nodes & links is replaced by the language of
frames
& slots.
6.
Define instance frames or instances.
Ans:
Frames which describe individual things are called instances.
7.
Frames which describe entire classes are called _____________.
Ans:
class frames or classes.
8. Is-a
slot - Define.
Ans:
Is-a slot (is-a-member-of-the-class) is a special slot which determines the
class
to which an instance frame belongs.
9. Ako
slot –Define.
Ans:
Ako slot (a-kind-of) is a special slot which ties classes together. It define
hierarchy
of class frames.
10. If
to traverse only one Ako slot then the frame to which it points is called
as_________
else if more than one then it is called as_________.
Ans:
direct subclass, subclass
11. If
to traverse only one Ako slot then the frame from which it emanates is called
as_________
else if more than one then it is called as_________.
Ans:
direct superclass, superclass
12.
_________ used to make and manipulate instances and classes.
Ans:
Access procedures
13.
Give some of the access procedures.
Ans:
Class constructor, instance constructor, slot writer, slot reader.
14.
_________ can make class frames that contain other slots and more than one
direct
superclass.
Ans:
Class constructor
15.
Instance constructor makes ________ frames. Its input consists of _________ and
its
output is _________.
Ans:
instance, the name of the class to which the instance belongs, an instance
that
belongs to those classes
16.
What is the job of slot writer?
Ans: It
installs slot values. Its input is a frame, the name of a slot, and a value to
be
installed.
17.
What is the job of slot reader?
Ans: It
retrieves slot values. Its input is a frame and the name of a slot; its output
is the
corresponding slot value.
18. The
slots in an instance are determined by that instance’s superclasses. If a
superclass
has a slot, then the instance is said to ________ that slot.
Ans:
inherit
19.
Inheritance enables ___________ procedures to move default slot values from
classes
to instances.
Ans:
when-constructed
20. The
expectations established by when-constructed procedures are called _______.
Ans:
defaults.
21. If
we have only one Ako slot exiting it is easy to form an ordered list called
________.
Ans:
class-precedence list
22. If
we have more than one Ako slot exiting then we need to use ___________ and
if it
also has more than one Is-a slot then we need to include _________.
Ans:
exhaustive depth-first ,left-to-right search procedures & up-to-join
proviso.
Topological-sorting.
23.
Explain exhaustive depth-first search.
Ans: It
explores all paths depth first, until each path reaches either a leaf node or
a
previously encountered node.
24.
Explain about up-to-join proviso.
Ans:Any
class that is encountered more than once during depth-first ,left-to-right
search
is ignored until that class is encountered for the last time.
25. How
to fill slots in a new instance?
Ans:
Compute class-precedence list using topological-sorting. Collect all
when-constructed
procedures for that slot, move along class-precedence list & use
the
most specific when-constructed procedure.
26. How
to compute an instance’s class-precedence list?
Ans:
Create fish-hook pairs. Until all fish-hook pairs are eliminated find the
exposed
classes, select the exposed class that is a direct superclass of the
lowestprecedence
class
on the emerging class precedence list, add it to the emerging
class
precedence list, strike all fish-hook pairs that contain the newly added class.
27.
List some of the demon procedures. Why it is called so?
Ans:
When-requested procedures, when-read procedures or when-written
procedures
are demon procedures. B’coz they lurk about doing nothing unless
they
see the request, read, or write operations they were designed to look for.
28.
____________ Procedures override slot values.
Ans:
When-requested
29.
____________ Procedures can maintain constraints.
Ans:
When-read and when-written
30.
____________ Procedures deal with perspectives and contexts.
Ans:
When-respect-to
31.
With no demons frame system can be viewed as _________.
Ans:
semantic nets
32.
Inheritance and demons introduce ________ semantics and
incorporates_________
knowledge.
Ans:
Procedural, procedural
33.
Object-oriented programming focuses on _________ knowledge.
Ans:
Shared
34.
____________ Procedure helps to perform an action in a manner suited to the
object
acted on.
Ans:
When-applied
UNIT-4
1. The
relational decision depends on -------
a)
relative importance of various goals.
b)
degree to which, they will be achieved.
1) ' a
' only 2) ' b' only 3) both a and b.
2.
Probabilities between 0 and 1 correspond to----------
a)
unequivocal belief that the sentence is false.
b)
unequivocal belief that the sentence is true.
c)
intermediate degrees of belief in the truth of the sentence.
3. As
the agent receives new percepts, its probability assesments are updated to
reflect the
new
evidence. Before the evidence is obtained , we talk about-----------
a)
prior probability
b)
posterior probability
c)
uncondotional probability
1) a
only 2) b only 3) c only 4) both a and c
4.
----------- says that every state has a degree of usefulness, or utility, to an
agent , and
that
the agent will prefer states with higher utility.
a)
utility theory.
b)
probability theory
c)
decision theory
5.----------
says that an agent is rational if and if only if it chooses the action that
yields
the
highest expected utility , averaged over all the possible outcomes of the
action.
a)
utility theory.
b)
probability theory
c)
Principal of Maximum Expected utility.
6. Any
notation for describing degrees of belief must be able to deal with
-----------issues.
a)nature
of the sentences to which the degrees of belief are assigned
b) the
dependence of the degree of belief on the agent's state of knowledge
c)both
a and b
7.----------
denotes a vector of probabilities of all the possible values of a random
variable
a)domain
b)probability
distribution
c)product
rule
8.Conditional
probabilities can be defined in terms of unconditional probabilities.(T / F).
9.------------is
an assignment of particular values to all the variables i.e. it completes
specification
of the state of domain.
a)joint
probability distribution
b)atomic
event
c)normalization
10.The
------------ view describes probabilities as a way of characterizing an agent's
beliefs,rather
than having any external physical significance.
a)subjectivist
b)objectivist
c)frequentist
12.------------data
structure is used to represent the dependence between variables and to
give a
concise specification of the joint probability distribution.
a)probabilistic
network
b)causal
network
c)belief
network
d)all
the three
13.The
main advantage of probabilistic reasoning over logical reasoning is in
-----------
14.
A---------- is a possible combination of values for the parent nodes.
a)conditioning
case
b)D_separation
c)leak
node
15.In
general, a conditional probability table for a Boolean variable with n Boolean
parents
contains ----------- independently specifiable probabilities.
a)n ²
b)2^n
c)n+2
16.The
two ways of understanding the semantics of belief networks
are---------&----------
17.Belief
networks contain no redundant probability values , except perhaps for one entry
in each
row of each conditional probability table.(T / F)
18.In a
------------ each subcomponent interacts directly with only a bounded number of
other
components.
a)locally
structured system
b)globally
structured system
c)acyclic
system
19.A
---------- node is the value specified exactly by the values of its parents
with no
uncertainity.
20.The
basic task for any probabilistic inference system is to compute the posterior
probability
distribution for a set of query variables , given exact values for some
evidence
variables.(T
/ F)
21.Belief
network can perform --------- to understand which aspects of the model have the
greatest
impact on the probabilities of the query variables.
a)causal
inferences
b)sensitivity
analysis
c)diagnostic
inferences
22.----------
methods transform the network into a probabilistically equivalent(but
topologically
different)polytree by merging offending nodes.
a)clustering
b)conditional
methods
c)neither
a nor b
23.----------methods
ues the network to generate a large number of concrete models of the
domain.
24.In
the general case , exact inference in belief networks is known to be NP_hard.(T
/ F)
25.----------method
transforms the network into simpler polytrees.
a)cutset
conditioning
b)clustering
methods
c)logic
sampling
26.A
set of variables that can be instantiated to yield polytrees is called--------
a)cutset
b)meganode
c)bounded
cutest
27.First
order logic exhibits strict monotonicity.(T / F)
28.-----------reasoning
says that the set of beliefs doesn’t grow monotonically over time
as new
evidence arrives.
29.-----------users
interval-valued degrees of belief to represent an agent’s knowledge of
the
probability of a proposition.
a)fuzzy
logic
b)Dempster-shafer
theory
c)rule
based approach
30.Belief
function(Bel(X)) computes the probability that the evidence supports the
proposition.(T
/ F)
31.-------------takes
a complex sentence and determines its truth value as a function of
truth
values of its components.
32.An
agent’s preferences between world states are captured by a---------, which
assigns
a
single number to express the desirability of a state.
33.The
axioms of utility theory are----------
34.The
difference between the expected monetary value of a lottery and its certainty
equivalent
is called------------
35.The
basic approach adapted in multiattribute utility theory is to identify
regularities in
the
preference behaviour we would expect to see.(T / F)
36.A
set of attributes is ------------ if each subset is utility independent of the
remaining
attributes.
37.------------networks
are a natural extension of belief networks containing decision and
utility
nodes in addition to chance nodes.
38.The
problem of calculating an optimal policy in an accessible , stochastic
environment
with a
known transition model is called a------------
39.------------&--------------
are two methods for calculating optimal policies , closely
related
to the general computational technique of dynamic programming.
40.Decomposing
the state into a set of state variables simplifies the handling of
inaccessible
environments.(T / F)
ANSWERS:
1) 3
2) 3
3) 4
4)a
5)c
6)c
7)b
8) T
9) a
10)a
12)d
13)allowing
the agent to reach rational decisions
14)a
15)b
16)as a
representation of joint probability distribution and as an encoding of
collection of
conditional
independence statements.
17)T
18)a
19)deterministic
20)T
21)b
22)a
23)stochastic
simulation
24)T
25)a
26)a
27)T
28)nonmonotonicity
29)b
30)T
31)fuzzy
logic
32)utility
function
33)orderability,transitivity,continuity,substituitability,monotonicity,decomposability
34)insurance
premium
35)T
36)mutually
utility independent
37)decision
38)Markov
decisionproblem
39)value
iteration & policy iteration
40)T
UNIT -
5
Planning:
1. The
planning agent and the problem solving agent differ in ________ of the
following.
a.representation
of goals and states
b.representation
of actions
c.representation
& construction of action sequences
d.all
2.
______can make a direct connection between states and actions.
a.Problem
solving agent
b.Planning
agent
c.both
d.none
3.
Which of the following are true about planning agent?
1.has
an “open up” reep of states,goals and actions.
2. free
to add actions to the plan wherever needed,rather in an
incremental
sequence
3.uses
the fact that “most parts of the world areindependent of most
other
parts.
a.1,3
b.1,2
c.all
d.none
4. The
idea behind situation calculus is ________.
a.the
current situation plays a vital role in deciding the next action or state
b.
actions that are taken would lead to a state transformation
c.both
d.none
5.
Operators of a planner consists of ___no. of components,namely,
a. 3 -
action description,precondition,effect
b.
2-action description,effect
c.
1-action description
d.
2-action ,resulting state
6. A
situation space planner means
a.one
which searches all the possible situations
b.one
which searches the available knowledge base and tries to
find a
similar situation,to carry out an action
c. one
which has no idea about the situation in which it is,but plans
randomly
to get into another situation
d. one
which tries out all the possible situations and selects the
optimal
one from them finally.
7.
Progression planner is
a. one
which is used for research purposes to progress the field of
planners
further.
b. One
which searches forward from the initial situation to get the
goal
situation
c.
Another name for situation space planner
d. none
of the above
8.
Regression planning is
1.searching
backwards from goal state to initial state
2.a way
to cut down the branching factor in the progression
planning
3.complicated
because there may be a conjunction of goals to be
achieved
a. 1,3
b. 1,2
c. all
d. none
9.Partial
plan is
a. a simple,initial,incomplete
plan
b. used
in a planning where the space of plans is searched rather than the
situations
to reach the goal
c. both
d. none
10.Refinement
operator means
a. an
ordinary operator refined by adding some extra constraints to it
b. the
one which takes a partial plan and adds constraints to it
c. a
special operator to search the KB effectively
d. an
operator to refine already available plan to get an optimized one.
11.Modification
operator is
a. the
one that is not a refinement operator
b. the
one used to modify a KB
c. the
one used to modify a plan
d. the
one used to modify a sample situation in KB to suit the current one
for
solving
12.
Principle of least commitment says,
a. it
is better to consider situations as least as possible
b. it
is better to make choices about things that you currently care about
c. it
is better to consider the plan that is used less frequently
d. none
13.Linearization(P)
means
a.
deriving a totally ordered plan from a plan P by adding constraints to it
b.
linearising the time to execute the plan P by certain methodologies
c. both
d. none
14. A
plan is a data structure of ____ of the following components.
1. set
of plans
2. set
of ordering constraints
3. set
of variable binding constraints
4. set
of causal links_
a.4-all
the above
b.
3-1,2,3
c.
2-1,3
d.3-1,2,4
15.
Solution can be thought as
a. a
plan that an agent can execute
b. that
guarantees achievement of the goal
c. a
complete_,consistent_ plan
d. all
16.
Threats to a causal link can be resolved by
a.
demotion®
b.
promotion®
c. both
d. none
17. To
deal the problems of incomplete and incorrect information_____ is/are
suitable
a.
conditional planning
b.
execution monitoring
c.both
d.none;these
kind of problems are not yet solved
18.
Conditional planning
a. plan
that is based on certain conditions on the situations for the
plan to
be executed
b. it
deals with incomplete information by constructing a conditional
plan
that accounts for each possible situation that could arise
c. both
d. none
19.
Execution monitoring
a.
monitoring what is happening while the agent executes a plan,the
agent
can tell when things go wrong
b.the
process of monitoring the execution of an agent while it
executes
an incomplete problem
c.both
d. none
20.
Conditional plan has
_
causal link is written as s1_s2 which means the purpose of s1 is to achieve a
precondition c of s2.
_ every
precondition of every step is achieved by some other step
_ there
are no contradictions in the ordering or binding constraints
® refer
to the figure in pg.353 of “AI,a modern approach “ by Russell,Norwick
a. a
condition that should be satisfied by all the actions that are to be
performed
b. a
step’s context which is simply the union of the conditions that
must
hold in order for the step to be executed
c. both
d. none
21.Action
monitoring
a.
checking the preconditions of each action as it is executed
b. the
process of monitoring the action taken by an agent in an
incomplete
situation
c. both
d. none
22. The
causes of plan failure can be due to
a.
bounded indeterminancy£
b.
unbounded indeterminancyV
c.both
d. none
LEARNING
The
various methods of learning are
1.Analysing
differences
- by
analyzing the differences thet appear in a sequence of
observations.
2.Explaining
experience
-
learning from experience by working exercises
3.
Correcting mistakes
- based
on how a procedure can repair previopusly acquired
knowledge
by exploiting its own errors.
4.
Recording cases
-
learning based on how it is possible to deal with problem
domains
in which good models are impossible to build.
5.Version
space method
-an
elegant way of learning using the positive and negative
examples
to build a version space which is used to guess what it
takes
to be a member of a class.
£
actions can have unexpected effects but the possible effects can be enumerated
and described as part of the action description axioim
V set
of possible unexrected outcomes is too large to be completely enumerated
6.Building
identification trees
-building
an identification tree by assembling the tests performed
in the
past.
-learning
by transforming the identification trees into perspicuous
set of
antecedent-consequent rules.
NOTE:
For
nueral networks, refer “ AI,a modern approach “ by Russell,Norwick.
ANSWERS:
1. d
2. b
3. c
4. b
5. a
6. a
7. b
8. c
9. c
10. b
11. a
12. b
13. a
14. a
15. d
16. c
17. c
18. b
19. a
20. b
21. a
22. c
