BCA (New) Assignment for the Session July 2010 6th Semester

BC005901 – Artificial Intelligence

1. What are the achievements of AI?

Ans. The achievements of AI are as follows: -

Deep thought is an international grand master chess player.

Sphinx can recognize continuous speech without training for each speaker. It operates in near

real time using a vocabulary of 1000 words and has 94% word accuracy.

Navlab is a truck that can drive along a road at 55 KMPH in normal traffic.

Carlton and United Breweries use an AI planning system to plan production of their beer.

Robots are used regularly in manufacturing.

Natural language interface to databases can be obtained on a PC.

Machine Learning methods have been used to build expert systems.

Expert systems are used regularly in finance, medicine, manufacturing, and agriculture.

2. Comment on “Heuristics are fallible.”

Ans. Heuristics are fallible because they rely on limited information, they may lead to a suboptimal

solution or to a dead end.

Heuristics is a rule of thumb or judgmental technique that leads to a solution some of the time

but provides no guarantee of success. It may in fact end in failure. Heuristics plays an important role

in search strategies because of the exponential nature of most problems. They help to reduce the

number of alternatives from an exponential number to a polynomial number and, thereby, obtain a

solution to a tolerable amount of time. When exhaustive search is impractical, it is necessary to

compromise for a constrained search which eliminates many paths but offers the promise of success

some of the time. Here, success may be considered to be finding an optimal solution a fair proportion

of the time or just finding good solutions much of the time.

3. What is the Drawback/Complexity of A*?

Ans. The catch with A* is that even though it is complete, optimally efficient, it still can’t always be

used, because for most problems, the number of nodes within the goal contour search space is still

exponential in the length of the solution.

Similarly to breadth first search, however the major difficulty with A* is the amount of space that

it uses.

4. What different kinds of knowledge need to be represented in AI?

Ans. Several kinds of information need to be represented in AI are as follows: -

Long –Term Knowledge: - This is accumulated knowledge about the world. It can include

simple data, general rules (every person has a mother), programs, and heuristic knowledge

(Knowledge of what is likely to work). The collection of long term knowledge is often called a

knowledge base (KB). Human long-term memory seems unlimited, but writing to it is slow.

Current Data: A representation of the facts of the current situation. Human short term

memory is very limited.

Conjectures: Courses of action or reasoning that are being considered but are not yet final.

5. Explain the process of Skolemization. How is this accomplished? Give suitable examples in

support of your answer.

Ans. Skolemization is the process of removing existential quantifiers by elimination. In the simple

translate into P (A), where A is a constant that does not appear elsewhere in the KB. But there

is the added complication that some of the existential quantifiers, even though move left, may still be

nested inside a universal quantifier.

Skolemization is accomplished as follows: -

If the first (leftmost) quantifier in an expression is an existential quantifier, replace all

occurrences of the variable it quantifier with an arbitrary constant not appearing

elsewhere and delete the quantifier. The same procedure should be followed for all other

existential quantifiers not preceded by a universal quantifier, in each case, using different

constant symbols in the substitution.

For each existential quantifier that is preceded by one or more universal quantifiers (is

within the scope of one or more universal quantifiers) replace all occurrences of the

existentially quantified variable by a function symbol not appearing elsewhere in the

expression. The argument assigned to the function should match all the variables

appearing in each universal quantifier which preceded the existential quantifier. This

existential quantifier should then be deleted. The same procedure should be repeated for

each remaining existential quantifier using a different function symbol and choosing

function arguments that correspond to all universally quantified variables that precede the

existentially quantified variable being replaced.

Example of Skolemization

Consider “Everyone has a heart”:

Λ Has (x, y)

If we just replaced y with a constant, H, we would get,

Λ Has (x, H)

Which says that everyone has the same heart H.? We need to say that the heart they have is not

necessarily shared, that is, it can be found by applying to each person a function that maps from person

to heart:

Λ Has (x, F(x))

Where F is a function name that does not appear elsewhere in the KB. F is called a Skolem

Function. In general, the existentially quantified variable is replaced by a term that consists of a Skolem

Function applied to all the variables universally quantified outside the existential quantifier in question.

Skolemization eliminates all existentially quantified variables, so we are now free to drop the universal

quantifiers, because any variable must be universally quantified.

6. Give the classification of different types of tasks of AI.

Ans. One possible classification of AI tasks is into 3 classes: Mundane tasks, Formal tasks and Expert

tasks.

Mundane Tasks

o Perception

o Vision

o Speech

o Natural Language understanding, generation and translation

o Common-sense Reasoning

o Simple reasoning and logical symbol manipulation

o Robot Control

Formal Tasks

o Games

Chess

Backgammon

Draughts

GO

o Mathematics

Geometry and Logic

Logic Theorist: - It proved mathematical theorems. It actually proved several

theorems from Classical Math Textbooks.

Integral Calculus

Programs such as Mathematical and Mathcad and perform complicated

symbolic integration and differentiation.

o Proving properties of Programs e.g. correctness

Expert Tasks

o Engineering

Design

Fault Finding

Manufacturing

Planning

Scientific Analysis

Medical Diagnosis

7. Explain briefly the process of matching production rules against working memory.

Ans. Production systems may vary on the expressive power of conditions in production rules.

Accordingly, the pattern matching algorithm which collects production rules with matched conditions

may range from the naïve-trying all rules in sequence, stopping at the first match-to the optimized, in

which rules are “compiled” into a network of inter-related conditions.

The latter is illustrated by the RETE algorithm, designed by Charles L. Forgy in 1983, which is used

in a series of production systems, called OPS and originally developed at Carnegie Mellon University

culminating in OPS5 in the early eighties. OPS5 may be viewed as a full-fledged programming

language for production system programming.

8. Comment on “Best-First is a combination of depth first and breadth first searches.

Ans. Depth first is good because a solution can be found without computing all nods and breadth first

is good because it does not get trapped in dead ends. The best first search allows us to switch

between paths thus gaining the benefit of both approaches. At each step the most promising node is

chosen. If one of the nodes chosen generates nodes that are less promising it is possible to choose

another at the same level and in effect the search changes from depth to breadth. If on analysis these

are no better than this previously unexpanded node and branch is not forgotten and the search

method reverts to the descendants of the first choice and proceeds, backtracking as it were.

9. Explain the properties of knowledge representation systems.

Ans. The following properties should be possessed by a knowledge representation system: -

Representational Adequacy the ability to represent the required knowledge.

Inferential Adequacy the ability to manipulate the knowledge represented to produce new

knowledge corresponding to that inferred from the original.

Inferential Efficiency the ability to direct the inferential mechanisms into the most productive

directions by storing appropriate guides.

Acquisition Efficiency the ability to acquire new knowledge using automatic methods

wherever possible rather than reliance on human intervention.

10. Explain the different strategies for the selection of clauses to be resolved.

Ans. Many different strategies have been tried for selecting the clauses to be resolved. These

includes: -

Level saturation or two-pointer method: - The outer pointer starts at the negated conclusion: the

inner pointer starts at the first clause. The two clauses denoted by the pointers are resolved if

possible, with the result added to the end of the list of clauses. The inner pointer is incremented

to the next clause until it reaches the outer pointer; then the outer pointer is incremented and

the inner pointer is reset to the front. The two-pointer method is a breadth-first method that will

generate many duplicate clauses.

Set of Support: - One clause in each resolution step must be part of the negated conclusion or a

clause derived from it. This can be combined with the two-pointer method by putting the clauses

from the negated conclusion at the end of the list. Set-of-support keeps the proof process focused

on the theorem to be proved rather than trying to prove everything.

Unit Preference: Clauses are prioritized, with unit clauses preferred, or more generally, shorter

clauses preferred. Resolution with a unit clause makes the result smaller.

Linear Resolution: - One clause in each step must be the result of the previous step. This is a

depth-first strategy. It may be necessary to back up to a previous clause if no resolution with the

current clause is possible.