instance, is an ideal AI Problem. There is no formal algorithm for its
realization, i.e., given a starting and a goal state, one cannot say prior to
execution of the tasks the sequence of steps required to get the goal from the
starting state. Such problems are called the ideal AI problems. The well-
known water-jug problem [35], the Travelling Salesperson Problem (TSP)
[35], and the n-Queen problem [36] are typical examples of the classical AI
problems. Among the non-classical AI problems, the diagnosis problems and
the pattern classification problem need special mention. For solving an AI
problem, one may employ both AI and non-AI algorithms. An obvious
question is: what is an AI algorithm? Formally speaking, an AI algorithm
generally means a non-conventional intuitive approach for problem solving.
The key to AI approach is intelligent search and matching. In an intelligent
search problem / sub-problem, given a goal (or starting) state, one has to reach
that state from one or more known starting (or goal) states. For example,
consider the 4-puzzle problem, where the goal state is known and one has to
identify the moves for reaching the goal from a pre-defined starting state.
Now, the less number of states one generates for reaching the goal, the better
is the AI algorithm. The question that then naturally arises is: how to control
the generation of states. This, in fact, can be achieved by suitably designing
some control strategies, which would filter a few states only from a large
number of legal states that could be generated from a given starting /
intermediate state. As an example, consider the problem of proving a
trigonometric identity that children are used to doing during their schooldays.
What would they do at the beginning? They would start with one side of the
identity, and attempt to apply a number of formulae there to find the possible
resulting derivations. But they won’t really apply all the formulae there.
Rather, they identify the right candidate formula that fits there best, such that
the other side of the identity seems to be closer in some sense (outlook).
Ultimately, when the decision regarding the selection of the formula is over,
they apply it to one side (say the L.H.S) of the identity and derive the new
state. Thus they continue the process and go on generating new intermediate
states until the R.H.S (goal) is reached. But do they always select the right
candidate formula at a given state? From our experience, we know the answer
is “not always”. But what would we do if we find that after generation of a
few states, the resulting expression seems to be far away from the R.H.S of
the identity. Perhaps we would prefer to move to some old state, which is
more promising, i.e., closer to the R.H.S of the identity. The above line of
thinking has been realized in many intelligent search problems of AI. Some of
these well-known search algorithms are:
a) Generate and Test
b) Hill Climbing
c) Heuristic Search
d) Means and Ends analysis
(a) Generate and Test Approach: This approach concerns the
generation of the state-space from a known starting state (root) of the problem
and continues expanding the reasoning space until the goal node or the
terminal state is reached. In fact after generation of each and every state, the
generated node is compared with the known goal state. When the goal is
found, the algorithm terminates. In case there exist multiple paths leading to
the goal, then the path having the smallest distance from the root is preferred.
The basic strategy used in this search is only generation of states and their
testing for goals but it does not allow filtering of states.
(b) Hill Climbing Approach: Under this approach, one has to first
generate a starting state and measure the total cost for reaching the goal from
the given starting state. Let this cost be f. While f ≤ a predefined utility value
and the goal is not reached, new nodes are generated as children of the current
node. However, in case all the neighborhood nodes (states) yield an identical
value of f and the goal is not included in the set of these nodes, the search
algorithm is trapped at a hillock or local extrema. One way to overcome this
problem is to select randomly a new starting state and then continue the above
search process. While proving trigonometric identities, we often use Hill
Climbing, perhaps unknowingly.
(c) Heuristic Search: Classically heuristics means rule of thumb. In
heuristic search, we generally use one or more heuristic functions to determine
the better candidate states among a set of legal states that could be generated
from a known state. The heuristic function, in other words, measures the
fitness of the candidate states. The better the selection of the states, the fewer
will be the number of intermediate states for reaching the goal. However, the
most difficult task in heuristic search problems is the selection of the heuristic
functions. One has to select them intuitively, so that in most cases hopefully
it would be able to prune the search space correctly. We will discuss many of
these issues in a separate chapter on Intelligent Search.
(d) Means and Ends Analysis: This method of search attempts to
reduce the gap between the current state and the goal state. One simple way to
explore this method is to measure the distance between the current state and
the goal, and then apply an operator to the current state, so that the distance
between the resulting state and the goal is reduced. In many mathematical
theorem- proving processes, we use Means and Ends Analysis.
Besides the above methods of intelligent search, there exist a good
number of general problem solving techniques in AI. Among these, the most
common are: Problem Decomposition and Constraint Satisfaction.
Problem Decomposition: Decomposition of a problem means breaking
a problem into independent (de-coupled) sub-problems and subsequently sub-
problems into smaller sub-problems and so on until a set of decomposed sub-
problems with known solutions is available. For example, consider the
following problem of integration.
I= ∫ (x2 + 9x +2) dx,
which may be decomposed to
∫ (x2 dx) + ∫ (9x dx) + ∫ (2 dx) ,
where fortunately all the 3 resulting sub-problems need not be decomposed
further, as their integrations are known.
Constraint Satisfaction: This method is concerned with finding the
solution of a problem by satisfying a set of constraints. A number of
constraint satisfaction techniques are prevalent in AI. In this section, we
illustrate the concept by one typical method, called hierarchical approach for
constraint satisfaction (HACS) [47]. Given the problem and a set of
constraints, the HACS decomposes the problem into sub-problems; and the
constraints that are applicable to each decomposed problem are identified and
propagated down through the decomposed problem. The process of re-
decomposing the sub-problem into smaller problems and propagation of the
constraints through the descendants of the reasoning space are continued until
all the constraints are satisfied. The following example illustrates the principle
of HACS with respect to a problem of extracting roots from a set of
inequality constraints.
Example 1.2: The problem is to evaluate the variables X 1, X2 and X3 from
the following set of constraints:
{ X1 ≥ 2; X2 ≥3 ; X1 + X2 ≤ 6; X1 , X2 , X3 ∈ I }.
For solving this problem, we break the ‘ ≥’ into ‘>’ and ‘=’ and propagate the
sub-constraints through the arcs of the tree. On reaching the end of the arcs,
we attempt to satisfy the propagated constraints in the parent constraint and
reduce the constraint set. The process is continued until the set of constraints
is minimal, i.e., they cannot be broken into smaller sets (fig. 1.3).
There exists quite a large number of AI problems, which can be solved
by non-AI approach. For example, consider the Travelling Salesperson
Problem. It is an optimization problem, which can be solved by many non-AI
algorithms. However, the Neighborhood search AI method [35] adopted for
this problem is useful for the following reason. The design of the AI
algorithm should be such that the time required for solving the problem is a
polynomial (and not an exponential) function of the size (dimension) of the
problem. When the computational time is an exponential function of the
dimension of the problem, we call it a combinatorial exploration problem.
Further, the number of variables to be used for solving an AI problem should
also be minimum, and should not increase with the dimension of the
problem. A non-AI algorithm for an AI problem can hardly satisfy the above
two requirements and that is why an AI problem should be solved by an AI
approach.
{ X1 ≥ 2; X2 ≥3 ; X1 + X2 ≤ 6; X1 , X2 , X3 ∈ I }
X1 = 2 X1 > 2
{ X1 =2, X2 ≥3 ; { X1 =3, X2 ≥3 ;
X1 + X2 ≤ 6; X j ∈ I, ∀j} X1 + X2 ≤ 6; Xj ∈ I, ∀j}
X2 =3 X2 >3 X2 =3 X 2> 3
{X1 =2, X2 =3} { X1 =2, X2 =4} {X1 =3, X2 =3} No solution
Fig. 1.3: The constraint tree, where the arcs propagate the constraints, and
the nodes down the tree hold the reduced set of constraints.
1.4 The Disciplines of AI
The subject of AI spans a wide horizon. It deals with the various kinds
of knowledge representation schemes, different techniques of intelligent
search, various methods for resolving uncertainty of data and knowledge,
different schemes for automated machine learning and many others. Among
the application areas of AI, we have Expert systems, Game-playing, and
Theorem-proving, Natural language processing, Image recognition, Robotics
and many others. The subject of AI has been enriched with a wide discipline
of knowledge from Philosophy, Psychology, Cognitive Science, Computer
Science, Mathematics and Engineering. Thus in fig. 1.4, they have been
referred to as the parent disciplines of AI. An at-a-glance look at fig. 1.4 also
reveals the subject area of AI and its application areas.
PARENT DISCIPLINES OF AI
.
Philosophy Maths. Psychology Computer
& Cog. Sc. Science
Artificial
Intelligence
* Reasoning * Learning * Planning * Perception
* Knowledge acquisition * Intelligent search
* Uncertainty management *Others
Subjects covered under AI
Game Theorem Language & Image Robotics &
Playing Proving Understanding Navigation
APPLICATION AREAS OF AI
Fig. 1.4: AI, its parent disciplines and application areas.
1.4.1 The Subject of AI
The subject of AI was originated with game-playing and theorem-proving
programs and was gradually enriched with theories from a number of parent
disciplines. As a young discipline of science, the significance of the topics
covered under the subject changes considerably with time. At present, the
topics which we find significant and worthwhile to understand the subject are
outlined below:
Tongue position adjustment
Motor Nerve
BRAIN
Voice System of
Learning System of
the Child the Child
Child’s pronunciation
Auditory
_ Nerve
Voice System of the
+ Hearing System
Mother
of the Child
Mother’s pronunciation
Fig. 1. 5: Pronunciation learning of a child from his mother.
Learning Systems: Among the subject areas covered under AI, learning
systems needs special mention. The concept of learning is illustrated here
with reference to a natural problem of learning of pronunciation by a child
from his mother (vide fig. 1.5). The hearing system of the child receives the
pronunciation of the character “A” and the voice system attempts to imitate it.
The difference of the mother’s and the child’s pronunciation, hereafter
called the error signal, is received by the child’s learning system through the
auditory nerve, and an actuation signal is generated by the learning system
through a motor nerve for adjustment of the pronunciation of the child. The
adaptation of the child’s voice system is continued until the amplitude of the
error signal is insignificantly low. Each time the voice system passes through
an adaptation cycle, the resulting tongue position of the child for speaking
“A” is saved by the learning process.
The learning problem discussed above is an example of the well-known
parametric learning, where the adaptive learning process adjusts the
parameters of the child’s voice system autonomously to keep its response
close enough to the “sample training pattern”. The artificial neural networks,
which represent the electrical analogue of the biological nervous systems, are
gaining importance for their increasing applications in supervised (parametric)
learning problems. Besides this type, the other common learning methods,
which we do unknowingly, are inductive and analogy-based learning. In
inductive learning, the learner makes generalizations from examples. For
instance, noting that “cuckoo flies”, “parrot flies” and “sparrow flies”, the
learner generalizes that “birds fly”. On the other hand, in analogy-based
learning, the learner, for example, learns the motion of electrons in an atom
analogously from his knowledge of planetary motion in solar systems.
Knowledge Representation and Reasoning: In a reasoning
problem, one has to reach a pre-defined goal state from one or more given
initial states. So, the lesser the number of transitions for reaching the goal
state, the higher the efficiency of the reasoning system. Increasing the
efficiency of a reasoning system thus requires minimization of intermediate
states, which indirectly calls for an organized and complete knowledge base.
A complete and organized storehouse of knowledge needs minimum search to
identify the appropriate knowledge at a given problem state and thus yields
the right next state on the leading edge of the problem-solving process.
Organization of knowledge, therefore, is of paramount importance in
knowledge engineering. A variety of knowledge representation techniques are
in use in Artificial Intelligence. Production rules, semantic nets, frames, filler
and slots, and predicate logic are only a few to mention. The selection of a
particular type of representational scheme of knowledge depends both on the
nature of applications and the choice of users.
Example 1. 3: A semantic net represents knowledge by a structured
approach. For instance, consider the following knowledge base:
Knowledge Base: A bird can fly with wings. A bird has wings. A bird has
legs. A bird can walk with legs.
The bird and its attributes here have been represented in figure 1.6 using a
graph, where the nodes denote the events and the arcs denote the relationship
between the nodes.
with
has
Fly A Bird Wings
can
has
Walk Legs
can with
Fig. 1.6: A semantic net representation of "birds".
Planning: Another significant area of AI is planning. The problems of
reasoning and planning share many common issues, but have a basic
difference that originates from their definitions. The reasoning problem is
mainly concerned with the testing of the satisfiability of a goal from a given
set of data and knowledge. The planning problem, on the other hand, deals
with the determination of the methodology by which a successful goal can be
achieved from the known initial states [1]. Automated planning finds
extensive applications in robotics and navigational problems, some of which
will be discussed shortly.
Knowledge Acquisition: Acquisition (Elicitation) of knowledge is
equally hard for machines as it is for human beings. It includes generation of
new pieces of knowledge from given knowledge base, setting dynamic data
structures for existing knowledge, learning knowledge from the environment
and refinement of knowledge. Automated acquisition of knowledge by
machine learning approach is an active area of current research in Artificial
Intelligence [5], [20].
Intelligent Search: Search problems, which we generally encounter in
Computer Science, are of a deterministic nature, i.e., the order of visiting the
elements of the search space is known. For example, in depth first and breadth
first search algorithms, one knows the sequence of visiting the nodes in a tree.
However, search problems, which we will come across in AI, are
non-deterministic and the order of visiting the elements in the search space is
completely dependent on data sets. The diversity of the intelligent search
algorithms will be discussed in detail later.
Logic Programming: For more than a century, mathematicians and
logicians were used to designing various tools to represent logical statements
by symbolic operators. One outgrowth of such attempts is propositional
logic, which deals with a set of binary statements (propositions) connected by
Boolean operators. The logic of propositions, which was gradually enriched to
handle more complex situations of the real world, is called predicate logic.
One classical variety of predicate logic-based programs is Logic Program
[38]. PROLOG, which is an abbreviation for PROgramming in LOGic, is a
typical language that supports logic programs. Logic Programming has
recently been identified as one of the prime area of research in AI. The
ultimate aim of this research is to extend the PROLOG compiler to handle
spatio-temporal models [42], [20] and support a parallel programming
environment [45]. Building architecture for PROLOG machines was a hot
topic of the last decade [24].
Soft Computing: Soft computing, according to Prof. Zadeh, is “an
emerging approach to computing, which parallels the remarkable ability of the
human mind to reason and learn in an environment of uncertainty and
imprecision” [13]. It, in general, is a collection of computing tools and
techniques, shared by closely related disciplines that include fuzzy logic,
artificial neural nets, genetic algorithms, belief calculus, and some aspects of
machine learning like inductive logic programming. These tools are used
independently as well as jointly depending on the type of the domain of
applications. The scope of the first three tools in the broad spectrum of AI is
outlined below.
♦ Fuzzy Logic: Fuzzy logic deals with fuzzy sets and logical connectives
for modeling the human-like reasoning problems of the real world. A
fuzzy set, unlike conventional sets, includes all elements of the universal
set of the domain but with varying membership values in the interval
[0,1]. It may be noted that a conventional set contains its members with a
value of membership equal to one and disregards other elements of the
universal set, for they have zero membership. The most common operators
applied to fuzzy sets are AND (minimum), OR (maximum) and negation
(complementation), where AND and OR have binary arguments, while
negation has unary argument. The logic of fuzzy sets was proposed by
Zadeh, who introduced the concept in systems theory, and later extended it
for approximate reasoning in expert systems [45]. Among the pioneering
contributors on fuzzy logic, the work of Tanaka in stability analysis of
control systems [44], Mamdani in cement kiln control
[19] , Kosko [15] and Pedrycz [30] in fuzzy neural nets, Bezdek in pattern
classification [3], and Zimmerman [50] and Yager [48] in fuzzy tools and
techniques needs special mention.
♦ Artificial Neural Nets: Artificial neural nets (ANN) are electrical
analogues of the biological neural nets. Biological nerve cells, called
neurons, receive signals from neighboring neurons or receptors through
dendrites, process the received electrical pulses at the cell body and
transmit signals through a large and thick nerve fiber, called an axon. The
electrical model of a typical biological neuron consists of a linear
activator, followed by a non-linear inhibiting function. The linear
activation function yields the sum of the weighted input excitation, while
the non-linear inhibiting function attempts to arrest the signal levels of the
sum. The resulting signal, produced by an electrical neuron, is thus
bounded (amplitude limited). An artificial neural net is a collection of
such electrical neurons connected in different topology. The most common
application of an artificial neural net is in machine learning. In a learning
problem, the weights and / or non-linearities in an artificial neural net
undergo an adaptation cycle. The adaptation cycle is required for updating
these parameters of the network, until a state of equilibrium is reached,
following which the parameters no longer change further. The ANN
support both supervised and unsupervised types of machine learning. The
supervised learning algorithms realized with ANN have been successfully
applied in control [25], automation [31], robotics [32] and computer
vision [31]. The unsupervised learning algorithms built with ANN, on the
other hand, have been applied in scheduling [31], knowledge acquisition
[5], planning [22] and analog to digital conversion of data [41].
♦ Genetic Algorithms: A genetic algorithm (GA) is a stochastic
algorithm that mimics the natural process of biological evolution [35]. It
follows the principle of Darwinism, which rests on the fundamental belief
of the “survival of the fittest” in the process of natural selection of
species. GAs find extensive applications in intelligent search, machine
learning and optimization problems. The problem states in a GA are
denoted by chromosomes, which are usually represented by binary strings.
The most common operators used in GA are crossover and mutation. The
processes of execution of crossover and mutation are illustrated in fig.1.7
and 1.8 respectively. The evolutionary cycle in a GA consists of the
following three sequential steps [23].
a) Generation of population (problem states represented
by chromosomes).
b) Genetic evolution through crossover followed by
mutation.
c) Selection of better candidate states from the generated
population.
In step (a) of the above cycle, a few initial problem states are first
identified. The step (b) evolves new chromosomes through the process of
crossover and mutation. In step (c ) a fixed number of better candidate states
are selected from the generated population. The above steps are repeated a
finite number of times for obtaining a solution for the given problem.
1110 011 X 1001 010 Parent chromosomes
crossover points
1110 010 1001 011
Offsprings obtained by crossover
Fig.1.7: Exchange of genetic information by crossover operation.
1 10010 110
randomly selected bit of mutation
1 0 0010110
mutated (complemented) bit
Fig. 1. 8: The mutation operation: randomly selected
bits are complemented.
Manage ment of Impre cision and Unce rtainty: Data and knowledge-
bases in many typical AI problems, such as reasoning and planning, are often
contaminated with various forms of incompleteness. The incompleteness of
data, hereafter called imprecision, generally appears in the database for i)
lack of appropriate data, and ii) poor authenticity level of the sources. The
incompleteness of knowledge, often referred to as uncertainty, originates in
the knowledge base due to lack of certainty of the pieces of knowledge.
Reasoning in the presence of imprecision of data and uncertainty of
knowledge is a complex problem. Various tools and techniques have been
devised for reasoning under incomplete data and knowledge. Some of these
techniques employ i) stochastic ii) fuzzy and iii) belief network models [16].
In a stochastic reasoning model, the system can have transition from one
given state to a number of states, such that the sum of the probability of
transition to the next states from the given state is strictly unity. In a fuzzy
reasoning system, on the other hand, the sum of the membership value of
transition from the given state to the next state may be greater than or equal to
one. The belief network model updates the stochastic / fuzzy belief assigned
to the facts embedded in the network until a condition of equilibrium is
reached, following which there would be no more change in beliefs. Recently,
fuzzy tools and techniques have been applied in a specialized belief network,
called a fuzzy Petri net, for handling both imprecision of data and
uncertainty of knowledge by a unified approach [14].
1.4.2 Applications of AI Techniques
Almost every branch of science and engineering currently shares the tools and
techniques available in the domain of AI. However, for the sake of the
convenience of the readers, we mention here a few typical applications, where
AI plays a significant and decisive role in engineering automation.
Expert Systems: In this example, we illustrate the reasoning process
involved in an expert system for a weather forecasting problem with special
emphasis to its architecture. An expert system consists of a knowledge base,
database and an inference engine for interpreting the database using the
knowledge supplied in the knowledge base. The reasoning process of a typical
illustrative expert system is described in Fig. 1.9. PR 1 in Fig. 1.9 represents
i-th production rule.
The inference engine attempts to match the antecedent clauses (IF parts)
of the rules with the data stored in the database. When all the antecedent
clauses of a rule are available in the database, the rule is fired, resulting in
new inferences. The resulting inferences are added to the database for
activating subsequent firing of other rules. In order to keep limited data in the
database, a few rules that contain an explicit consequent (THEN) clause to
delete specific data from the databases are employed in the knowledge base.
On firing of such rules, the unwanted data clauses as suggested by the rule are
deleted from the database.
Here PR1 fires as both of its antecedent clauses are present in the
database. On firing of PR1, the consequent clause “it-will-rain” will be added
to the database for subsequent firing of PR2.
Database
Knowledge Base
PR1: if (it-is-hot) and It-is-
(the-sky-is-cloudy) the-sky-is-cloudy
hot then (it-will-rain).
PR2: if (it-rains) Search
then (road-at-Calcutta
-becomes-flooded)
Inference Engine
Inferences
it-will-rain
the-road-at-Calcutta-becomes-flooded
Fig. 1. 9: Illustrative architecture of an expert system.
Image Understanding and Computer Vision: A digital image
can be regarded as a two-dimensional array of pixels containing gray levels
corresponding to the intensity of the reflected illumination received by a
video camera [6]. For interpretation of a scene, its image should be passed
through three basic processes: low, medium and high level vision (fig.1.10).
The importance of low level vision is to pre-process the image by filtering
from noise. The medium level vision system deals with enhancement of
details and segmentation (i.e., partitioning the image into objects of interest
). The high level vision system includes three steps: recognition of the
objects from the segmented image, labeling of the image and interpretation of
the scene. Most of the AI tools and techniques are required in high level
vision systems. Recognition of objects from its image can be carried out
through a process of pattern classification, which at present is realized by
supervised learning algorithms. The interpretation process, on the other hand,
requires knowledge-based computation.
Navigational Planning for Mobile Robots: Mobile robots, sometimes
called automated guided vehicles (AGV), are a challenging area of research,
where AI finds extensive applications. A mobile robot generally has one or
more camera or ultrasonic sensors, which help in identifying the obstacles on
its trajectory. The navigational planning problem persists in both static and
dynamic environments. In a static environment, the position of obstacles is
fixed, while in a dynamic environment the obstacles may move at arbitrary
directions with varying speeds, lower than the maximum speed of the robot.
Many researchers using spatio-temporal logic [7-8] have attempted the
navigational planning problems for mobile robots in a static environment. On
the other hand, for path planning in a dynamic environment, the genetic
algorithm [23], [26] and the neural network-based approach [41], [47] have
had some success. In the near future, mobile robots will find extensive
applications in fire-fighting, mine clearing and factory automation. In accident
prone industrial environment, mobile robots may be exploited for automatic
diagnosis and replacement of defective parts of instruments.
Camera Low level vision
Pre-processing Enhancement
Labeling Recognition Segmentation
Medium level vision
Interpretation
High level vision
high level inferences
Fig. 1.10: Basic steps in scene interpretation.
Speech and Natural Language Unde r standing: Understanding
of speech and natural languages is basically two class ical probl ems. In
speech analysis, the main probl em is to separate the syllables of a spoken
word and determine features like ampli tude, and fundamental and harmonic
frequ encies of each syllable. The words then could be ident ified from the
extracted featu res by pattern class ification techn iques. Recen tly, artificial
neural networks have been employed [41] to class ify words from their
features. The probl em of understanding natural languages like Engli sh, on
the other hand, includes syntactic and semantic interpretation of the words in
a sentence, and sentences in a paragraph. The syntactic steps are required to
analyze the sentences by its grammar and are similar with the steps of
compilation. The semantic analysis, which is performed following the
syntactic analysis, determines the meaning of the sentences from the
association of the words and that of a paragraph from the closeness of the
sentences. A robot capable of understanding speech in a natural language will
be of immense importance, for it could execute any task verbally
communicated to it. The phonetic typewriter, which prints the words
pronounced by a person, is another recent invention where speech
understanding is employed in a commercial application.
Scheduling: In a scheduling problem, one has to plan the time schedule of
a set of events to improve the time efficiency of the solution. For instance in
a class-routine scheduling problem, the teachers are allocated to different
classrooms at different time slots, and we want most classrooms to be
occupied most of the time. In a flowshop scheduling problem [42], a set of
jobs J1 and J2 (say) are to be allocated to a set of machines M 1, M 2 and M 3.
(say). We assume that each job requires some operations to be done on all
these machines in a fixed order say, M 1, M 2 and M 3. Now, what should be
the schedule of the jobs (J1-J2) or (J2 –J1), so that the completion time of both
the jobs, called the make-span, is minimized? Let the processing time of jobs
J1 and J2 on machines M 1, M 2 and M 3 be (5, 8, 7) and (8, 2, 3) respectively.
The gantt charts in fig. 1.11 (a) and (b) describe the make-spans for the
schedule of jobs J1 - J2 and J2 - J1 respectively. It is clear from these figures
that J1-J2 schedule requires less make-span and is thus preferred.
J1 J2
M1
5 8
J1 J2
M2
8 2
J1 J2
M3
7 3
make-span = 23
Job completion time
(a) The J1 - J2 schedule.
J2 J1
M1
8 5
J2 J1
M2
2 8
J2 J1
M3
3 7
make-span = 28
Job completion time
(b): The J2 - J1 schedule where the hatched lines indicate waiting
time of the machines.
Fig. 1.11: The Gantt charts for the flowshop scheduling problem
with 2 jobs and 3 machines.
Flowshop scheduling problems are a NP complete problem [1] and
determination of optimal scheduling (for minimizing the make-span) thus
requires an exponential order of time with respect to both machine-size and
job-size. Finding a sub-optimal solution is thus preferred for such scheduling
problems. Recently, artificial neural nets and genetic algorithms have been
employed to solve this problem. The heuristic search, to be discussed
shortly, has also been used for handling this problem [34].
Intelligent Control: In process control, the controller is designed from the
known models of the process and the required control objective. When the
dynamics of the plant is not completely known, the existing techniques for
controller design no longer remain valid. Rule-based control is appropriate in
such situations. In a rule-based control system, the controller is realized by a
set of production rules intuitively set by an expert control engineer. The
antecedent (premise) part of the rules in a rule-based system is searched
against the dynamic response of the plant parameters. The rule whose
antecedent part matches with the plant response is selected and fired. When
more than one rule is firable, the controller resolves the conflict by a set of
strategies. On the other hand, there exist situations when the antecedent part
of no rules exactly matches with the plant responses. Such situations are
handled with fuzzy logic, which is capable of matching the antecedent parts of
rules partially/ approximately with the dynamic plant responses. Fuzzy
control has been successfully used in many industrial plants. One typical
application is the power control in a nuclear reactor. Besides design of the
controller, the other issue in process control is to design a plant (process)
estimator, which attempts to follow the response of the actual plant, when
both the plant and the estimator are jointly excited by a common input signal.
The fuzzy and artificial neural network-based learning techniques have recently
been identified as new tools for plant estimation [25], [43].
1.5 A Brief History of AI
Professor Peter Jackson of the University of Edinburgh classified the history
of AI into three periods namely i) the classical period (of game playing and
theorem proving), ii) the romantic period, and iii) the modern period [12]; the
major research work carried out during these periods is presented below.
1.5.1 The Classical Period
This period dates back to 1950. The main research works carried out during
this period include game playing and theorem proving. The concept of state-
space approach for solving a problem, which is a useful tool for intelligent
problem-solving even now, was originated during this period [27].
The period of classical AI research began with the publication of
Shannon’s paper on chess (1950) [35] and ended with the publication by
Feigenbaum and Feldman [10]. The major area of research covered under this
period is intelligent search problems involved in game-playing and theorem-
proving. Turing’s “test”, which is a useful tool to test machine intelligence,
originated during this period.
1.5.2 The Romantic Period
The romantic period started from the mid 1960s and continued until the mid
1970s. During this period, people were interested in making machines
“understand”, by which they usually mean the understanding of natural
languages. Winograd’s (1972) SHRDLU system [46], a program capable of
understanding a non-trivial subset of English by representing and reasoning
about a restricted domain (a world consisting of toy blocks), in this regard
needs special mention. The knowledge representation scheme using special
structures like “semantic nets” was originated by Quillian [33] during this
period. Minisky (1968) also made a great contribution from the point of view
of information processing using semantic nets. Further, knowledge
representation formalisms using frames, which was another contribution of
Minisky during this period, also need special mention [28].
1.5.3 The Modern Period
The modern period starts from the latter half of the 1970s to the present day.
This period is devoted to solving more complex problems of practical
interest. The MYCIN experiments of Stanford University [4], [39] resulted in
an expert system that could diagnose and prescribe medicines for infectious
bacteriological diseases. The MECHO system for solving problems of
Newtonian machines is another expert system that deals with real life
problems. It should be added that besides solving real world problems,
researchers are also engaged in theoretical research on AI including heuristic
search, uncertainty modeling and non-monotonic and spatio-temporal
reasoning. To summarize, this period includes research on both theories and
practical aspects of AI.
1.6 Characteristic Requirements for the
Realization of the Intelligent Systems
The AI problems, irrespective of their type, possess a few common
characteristics. Identification of these characteristics is required for designing a
common framework for handling AI problems. Some of the well-known
characteristic requirements for the realization of the intelligent systems are
listed below.
1.6.1 Symbolic and Numeric Computation
on Common Platform
It is clear from the previous sections that a general purpose intelligent
machine should be able to perform both symbolic and numeric computations
on a common platform. Symbolic computing is required in automated
reasoning, recognition, matching and inductive as well as analogy-based
learning. The need for symbolic computing was felt since the birth of AI in
the early fifties. Recently, the connectionist approach for building intelligent
machines with structured models like artificial neural nets is receiving more
attention. The ANN based models have successfully been applied in learning,
recognition, optimization and also in reasoning problems [29] involved in
expert systems. The ANNs have outperformed the classical approach in many
applications, including optimization and pattern classification problems.
Many AI researchers, thus, are of the opinion that in the long run
the connectionist approach will replace the classical approach in all respects.
This, however, is a too optimistic proposition, as the current ANNs require
significant evolution to cope with the problems involved in logic
programming and non-monotonic reasoning. The symbolic and connectionist
approach, therefore, will continue co-existing in intelligent machines until the
latter, if ever, could replace the former in the coming years.
1.6.2 Non-Deterministic Computation
The AI problems are usually solved by state-space approach, introduced in
section 1.3. This approach calls for designing algorithms for reaching one or
more goal states from the selected initial state(s). The transition from one
state to the next state is carried out by applying appropriate rules, selected
from the given knowledge base. In many circumstances, more than one rule is
applicable to a given state for yielding different next states. This informally is
referred to as non-determinism. Contrary to the case, when only one rule is
applicable to a given state, this system is called deterministic. Generally AI
problems are non-deterministic. The issues of determinism and non-
determinism are explained here with respect to an illustrative knowledge-based
system. For instance, consider a knowledge base consisting of the following
production rules and database.
Production Rules
PR1: IF (A) AND (B) THEN ( C ).
PR2: IF ( C ) THEN ( D).
PR3: IF ( C ) AND ( E ) THEN (Y).
PR4: IF (Y) THEN (Z).
Database: A, B, E.
The graph representing the transition of states for the above reasoning
problem is presented in fig.1.12. Let A and B be starting states and Z be the
goal state. It may be noted that both PR2 and PR3 are applicable at state (C)
yielding new states. However, the application of PR3 at state (C) can
subsequently lead to the goal state Z, which unfortunately remains unknown
until PR4 is applied at state Y. This system is a typical example of non-
determinism. The dropping of PR2 from the knowledge base, however, makes
the system deterministic. One formal approach for testing determinism / non-
determinism of a reasoning system can be carried out by the following
principle:
A
C D
Z
AND
B
E AND Y
Fig. 1.12: A Petri-like net representing non-determinism in a reasoning system
with initial states A and B and goal state Z.
Principle for testing determinism: After deriving the goal state
from the initial states, continue marking (backtracking) the parents of each
node starting from the goal node, until the initial states are reached. If
unmarked nodes are detected, then the system is non-deterministic; otherwise
it is deterministic. It may be noted that testing of determinism in a
knowledge-based system, for any set of starting and goal states, is a distinct
problem and no conclusion about determinism can be drawn for modified
initial or goal states.
The principle for testing determinism in the proposed knowledge-based
system is illustrated here with reference to the dependence graph (Petri-like
net) of fig.1.12. It may be noted that while backtracking on the graph, node
D is not marked and thus the system is non-deterministic.
Besides reasoning, non-determinism plays a significant role in many
classical AI problems. The scope of non-determinism in heuristic search has
already been mentioned. In this section, we demonstrate its scope in
recognition problems through the following example.
Example 1.3: This example illustrates the differences of deterministic and
non-deterministic transition graphs [9], called automata. Let us first consider
the problem of recognition of a word, say, “robot”. The transition graph (fig.
1.13(a)) for the current problem is deterministic, since the arcs emerging out
from a given state are always distinct. However, there exist problems, where
the arcs coming out from a state are not always distinct. For instance,
consider the problem of recognizing the words “robot” and “root”. Here, since
more than one outgoing arc from state B (fig. 1.13(b)) contains the same label
(o), they are not distinct and the transition graph is non-deterministic.