CS308 : Artificial Intelligence

Introduction
• How we think, perceive, understand, predict, and manipulate.
• Different types and degrees of intelligence occur in people,
several animals, and a few machines.
• AI goes further - it attempts not just to understand but also to
build intelligent entities.
• McCarthy coined the term artificial intelligence in 1955, and
he created the computer programming language LISP in 1958
which was initially used primarily by the AI community.
• AI currently encompasses a huge variety of subfields.

 What is Problem?
• A problem is an obstacle that is troublesome to attain a
desired goal, or an issue that is unresolved.
• Each problem has an answer or solution.
• A procedure that makes the problem navigation
towards the goal is called solution.
• It can be algorithm or a hard-core implementation
to achieve the goal defined by the problem.

 Types of Problem and Solution

• Depending on the way in which the problems are
solved, problems are classified into the following
types:
1. Structured problem
2. Unstructured problem

 Structured problem
• Structural problems are the ones for which there exists a
specific algorithm to achieve the goal.
• The same algorithm is run against variety input data still
giving a guarantee of the problem being solved.
• Since the structure of the solution (that is in the algorithm)
remains the same, even if the input data changes, these
problems are called structured problems.
• For example-
– To sort the students according to their total marks, the details about
the students may be stored in the database, the sort algorithm will
then get the details of students from the database and will solve the
problem.

 Unstructured problem
• Unstructured problems are the problems for which there does
not exist a specific algorithm to achieve the goal.
• What step to take to achieve the goal depends on what is the
current state of the problem.
• AI is an attempt to make a computer to solve unstructured
problems.
• For example-
– A problem of playing chess or a problem to write a program to
perform heart surgery, etc., are unstructured problems, because there
does not exist any specific algorithm to solve it. Such problems are
solved using a Knowledge Base.

 What is AI?
• The study of mental facilities through the use of
computational models. - Charniak and McDermott, 1985.
• The study of how to make computer do things at which at the
moment, people are better. – Rich and Knight 1991
• The scientific understanding of the mechanisms underlying
thought and their intelligent behavior and their embodiment
in machines. - American Association of Artificial Intelligent
(AAAI)
• AI strives to understand and build intelligent entities (also
help us to learn more about ourselves – Russell and Norvig.
• AI is The science and engineering of making intelligent
machines, especially intelligent programs. - John McCarthy,
father of Artificial Intelligence.

 What is AI
• Top - thought
processes and
reasoning,
• Bottom - behaviour.
• Left - measure success
in terms of fidelity to
human performance.
• Right - measure
against an ideal
performance measure,
called rationality.
• A system is rational if
it does the “right
thing,” given what it
knows.

Acting humanly: The Turing Test approach

• The computer would need to possess the following
capabilities:
‣ Natural language processing - to communicate successfully
‣ Knowledge representation to store what it knows or hears;
‣ Automated reasoning to use the stored information to
answer questions and to draw new conclusions;
‣ Machine learning to adapt to new circumstances and to
detect and extrapolate patterns
• Computer vision to perceive objects
• Robotics to manipulate objects and move about.

Thinking humanly: The cognitive approach

• If a program thinks like a human, we must have
some way of determining how humans think.
• We need to get inside the actual workings of human
minds.
• Cognitive science brings together computer models
from AI and experimental techniques from
psychology to construct precise and testable
theories of the human mind.

Thinking rationally : laws of thought approach

• Greek philosopher Aristotle provided irrefutable reasoning

processes called syllogisms.
• Syllogisms provides patterns for argument structures that
always yielded correct conclusions when given correct
premises - Example.
• The laws of thought were supposed to govern the operation
of the mind, their study initiated the field called logic.
• The socalled logicist tradition within artificial intelligence
hopes to build on such programs to create intelligent
systems.

Acting rationally: The rational agent approach

• An agent is just something that acts (agent comes from the
Latin agere, to do).
• Computer agents are expected to do more than simple computer
programs - operate autonomously, perceive their environment,
persist over a prolonged time period, adapt to change, and
create and pursue goals.
• A rational agent is one that acts so as to achieve the best
outcome or, when there is uncertainty, the best expected
outcome.
• Limited rationality - acting appropriately when there is not
enough time to do all the computations one might like.

 History and Foundation of AI

1. 1943–1952 : Beginning (model of neuron , Turing Test, first neural network)
2. 1952–1969: Early enthusiasm (Sobering up, Game of Checkers, Logic
Theory, LISP, machine translation, Scaling up to larger problems)
3. 1956 : The birth of artificial intelligence
4. 1970–1979: Knowledge-based systems (DENDRAL : 450 rules, MYCIN :
550 rules, Natural Language Processing, PROLOG – Logical programming
language, Frame Theory for data structure to represent a situation)
5. 1980–2010: AI becomes an industry (several million to billion dollars, era
of Intelligent Agents, Machine Learning, Robotics.
6. 2010–till date: Era of Deep Learning (stunning advancement, availability of
large dataset)

Branches of AI
 AI - State of the Art
➢Game playing
➢Mathematics
➢Autonomous control
➢Diagnosis
➢ Logistics planning
➢Autonomous planning and scheduling
➢Language understanding and problem solving
➢Robotics
➢Natural Language Generation

 Applications of AI
➢Speech recognition
➢Virtual agents
➢Text analytics and NLP
➢Robotic process automation
➢Biometrics
➢Deep learning platforms
➢Decision management
➢AI-optimized hardware
➢Machine-learning platforms

 AI Programming languages
• A number of programming languages exist that are used to build AI
systems.
• Here are some languages that are most typically used for creating the
AI projects:
➢ PROLOG
➢ LISP
➢ R
➢ Python
➢ Java
➢ C++

Intelligent Agents
 Agents and Environments
• Intelligent Agent (IA) is a self sufficient substance which
watches and follows up on a situation and coordinates its
movement towards accomplishing objectives.
• An agent is anything that can be viewed as perceiving its
environment through sensors and acting upon that
environment through actuators.
• These may be extremely straight forward or extremely
complex.
• Small set of design principles for building successful
agents systems that can reasonably be called intelligent.

 Structure of Intelligent Agent

• The structure of agent can be represented as:
Agent= Architecture + Program
– Architecture : The machinery that an agent executes on
– Agent program: An implementation of an agent
function

 Agent terminology
1. Performance measure : It is the criteria determining the success of
an agent.
2. Action : It is the action performed by an agent after any specified
sequence of the percepts.
3. Percept : It is defined as an agent’s perceptual inputs at a specified
instance.
4. Percept sequence : It is defined as the history of everything that an
agent has perceived till date.
5. Agent function (F) : It is defined as a map from the precept
sequence to an action.
Agent function, a = F(p)
where p is the current percept, a is the action carried out, and F is
the agent function

 Agent terminology
• If P is the set of all precepts, and A is the set of all actions
then F maps precepts to actions.
F : P → A
• Generally, an action may be dependent of all the precepts
observed, not only the current percept,
ak = F( p0 p1 p2 … pk )
Where p0 , p1, p2, … ,pk is the sequence of percepts recorded
till date, ak is the resulting action carried out and F now
maps percept sequences to action
F : P * → A

 Vacuum cleaner Agent

• The goal of vacuum cleaner is to clean the environment.
• The environment has two locations of the room as square
A(left Square) and Square B(Right Square).
• Vacuum cleaner perceives in which square it is in and if there
is dirt in the square.
• It may choose to move left, right, suck up the dirt or do
nothing.
– Goal / Performance : All the rooms are well cleaned.
– Actions : Left, right, suck and no-op (Doing nothing).
– Percept : Location and status, for example, [A, Dirty].
– Agent function : Mapping of precept sequence to an action.

 Percept sequence Action

Vacuum [A, Clean] Right

cleaner Agent [A, Dirty] Suck

[B, Clean] Left
[B, Dirty] Suck
[A, Dirty], [A, Clean] Right
[A, Clean], [B, Dirty] Suck
[B, Dirty], [B, Clean] Left
[B, Clean], [A, Dirty] Suck
[A, Clean], [B, Clean] No-op
[B, Clean], [A, Clean] No-op
 Rational Agent
• What is rational at any given time depends on four things:
‣ The performance measure that defines the criterion of
success.
‣ The agent’s prior knowledge of the environment.
‣ The actions that the agent can perform.
‣ The agent’s percept sequence to date.
• A rational agent : For each possible percept sequence, a rational
agent should select an action that is expected to maximize its
performance measure, given the evidence provided by the
percept sequence and whatever built in knowledge the agent has.

 Agents and Environments

• The irrational agent
• Penalty
• A better agent - do nothing
• information gathering
• Learning
• Autonomy
• A rational autonomous agent

 The task environment (PEAS)

• Specify the task environment as fully as possible.
• All these are grouped under the heading of the task
environment called PEAS (Performance,
Environment, Actuators, Sensors) description.
• P, Performance : What the agent is expected to
achieve.
• E, Environment : What the agent is interacting with
• A, Actuators : The outputs of the AI system.
• S, Sensors : The inputs to the AI system.

 The task environment

• More complex problem: an automated taxi driver.
The full driving task is extremely open-ended.
 The task environment
• Fully vs. partially observable
• Single agent vs. multiagent
• Deterministic vs. stochastic
• Episodic vs. Sequential environments
• Static vs. dynamic
• Discrete vs. continuous
• Known vs. unknown
• The hardest case is partially observable, multiagent,
stochastic, sequential, dynamic, continuous, and unknown.

The task environment

 THE STRUCTURE OF AGENTS

• The agent program implements the agent function - the mapping from percepts to actions.
THE STRUCTURE OF AGENTS
 THE STRUCTURE OF AGENTS
• Four types of agent programs that embody the
principles underlying almost all intelligent
systems:
‣ Simple reflex agents
‣ Model-based reflex agents
‣ Goal-based agents
‣ Utility-based agents
‣ Learning agent

 Simple reflex agents

• The simplest kind of agent.
• Select actions on the basis of the current percept,
ignoring the rest of the percept history.
• Then triggers some established connection in the
agent program to perform the action

Simple
Reflex
Agents

 Limitations
• Intelligence level in these agents is very limited.
• It works only in a fully observable environment.
• It does not hold any knowledge or information of
non perceptual parts of state.
• Because of the static knowledge based, it is usually
too big to generate and store.
• If any change in the environment happens, the
collection of the rules are required to be updated.

 Model-based reflex agents

• One problem with the simple reflex agents is that their
activities are dependent of the recent data provided by their
sensors.
• More effective way to handle partial observability is to keep
track of the part of the world it can’t see now.
• The agent maintains some sort of internal state that depends on
the percept history and thereby reflects at least some of the
unobserved aspects of the current state.
• An agent that uses such a model is called a model-based agent.
• Model-based agent is known as Reflex agents with an internal
state.

Model
Based
Agents

 Goal-based agents
• Knowing something about the current state of the
environment is not always enough to decide what to do.
• The agent needs some sort of goal information that describes
situations the final destination.
• The agent program can combine this with the model based
reflex agent to choose actions that achieve the goal.
• Goal-based action selection may be straightforward or
tricky. Searching and planning are the area of study to find
action sequence that achieve agents goal.
• They are more adaptable and flexible.
• One can specify another goal rather than reprogramming all
the rules.

Goal-
based
agents
 Goal-based agents
• Goal-based agent appears less efficient but it is more flexible due to
the knowledge that supports its decisions is represented explicitly
and can be modified.
• This will automatically cause all of the relevant behaviors to be
altered to suit the new conditions.
• The goal-based agent’s behavior can easily be changed to go to a
different destination, simply by specifying that destination as the
goal.
• For the reflex agent, on the other hand, we would have to rewrite
many condition–action rules.
• The reflex agent’s rules for when to turn and when to go straight will
work only for a single destination; they must all be replaced to go
somewhere new.

 Utility based agent

• Goals individually are insufficient to produce top high-quality behavior.
• When there are conflicting goals, only some of which can be achieved (e.g
speed and safety), the utility function specifies the appropriate tradeoff.
• When there are several goals that the agent can aim for, none of which can
be achieved with certainty, utility provides a way in which the likelihood
of success can be weighed against the importance of the goals.
• Given proper criteria, it might be conceivable to pick ‘best’ sequence of
actions from a number that all result in the goal being achieved.
• Any utility-based agent can be depicted as having an utility capacity that
maps a state, or grouping of states, on to a genuine number that speaks to
its utility or convenience or usefulness.
• A rational utility-based agent chooses the action that maximizes the
expected utility of the action outcomes.

•
 Learning agent
• By actively exploring and experimenting with their environment, the most
powerful agents are able to learn.
• A learning agent can be further divided into the four conceptual components
• performance element - responsible for selecting external actions.
• learning element - responsible for making improvements
• problem generator - responsible for suggesting actions that will lead to
new and informative experiences
• Critic - Provides feedback on how agent is doing and determines how the
performance element should be modified to do in future.
• The performance element is what we have previously considered to be the
entire agent, it takes in percepts and decides on actions.
• The learning element uses feedback from the critic on how the agent is doing
and determines how the performance element should be modified to do
better in the future.

Learning agent
 Components of Agent Programs
• Representations along an axis of increasing
complexity and expressive power - atomic,
factored, and structured.
 Components of Agent Programs
• Atomic - Each state of the world is indivisible, it has no internal
structure.
• Factored - splits up each state into a fixed set of variables or
attributes, each of which can have a value.
• Two different atomic states have nothing in common and can share
some attributes (GPS location) and not others (fuel status), this
makes it much easier to work out how to turn one state into another.
• We would need a structured representation, in which objects and
their various and varying relationships can be described explicitly.
• Structured representations underlie relational databases and first-
order logic In fact, almost everything that humans express in natural
language concerns objects and their relationships.

 PROBLEM SOLVING
- Solving problem by searching

 Introduction
• In AI, the term problem solving is given to the analysis
of how computers can be made to find solutions in
restricted environments.
• Problem solving is defined as the way in which an
agent finds a sequence of actions that achieves its
goals, when no single action will do.
• Problem formulation requires abstracting real-world
details to define a state space that can be explored.
• This step of abstraction is performed by an agent
called as a problem-solving agent.
• Intelligent agents are supported to act in such a way
that the environment goes through a sequence of states
that maximises the performance measures.

 Problem Solving Agent

Problem solving agent is a form of goal based agent.
Consider a goal to be a set of world states in which a
goal is satisfied.
1. Goal state: A state that describes the objective that the
agent is trying to achieve is called as the goal state of
the agent.
2. Action: When there is a transition between the world
states an action is said to be performed.

 Problem Solving Agent

Steps in problem solving by problem solving agent
1. Goal formulation: On the basis of the current situation and
agent’s performance measure, it is the first step in problem
solving. Agent’s task is to find out sequence of actions that
will get it to the goal state. Goals help organise behaviour by
limiting the objectives the agent is trying to achieve.
2. Problem formulation: It is a process of deciding what actions
and states to consider given a goal.
3. Choosing the best sequence: 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 the states of known value and then choosing the best
sequence.

 Problem Solving Agent

• Problem formulation is the process of deciding what
actions and states to consider given a goal.
• It is defined by four component
‣ Initial State : The starting state of the agent.
‣ Successor Function : Function that gives the
description of all possible action available to the agent.
‣ Goal test : this determines the given state is goal state
or not.
‣ Path Cost : The function that assigns the numeric cost
to each path. The problem solving agent chooses a cost
function that reflects its own performance measures.

 Problem Solving Agent

• A simple problem solving goal based agent formulates a goal and a problem,
searches for a sequence of actions that would solve the problem, and then executes
the actions one at a time.
• When this is complete, it formulates another goal and starts over.

 State-space Representation
• State space is a tree which represents all the possible
states in the problem.
• It is complete set of states including start and goal
states, other valid states and transitions, where the
answer of the problem is to be searched.
• A state–space representation allows for the formal
definition of a problem, which makes the movement
from initial state to the goal state quite easily.
• State–space search is a process in AI, in which
successive configurations or states of an instance are
considered, with the goal of finding a goal state with a
desired property.

 State-space Representation
• AI problem can be solved using following steps:
‣ Describe states
‣ Identify initial state
‣ Identify set of rules (all possible action)
‣ Identify goal state
‣ Path cost
• When these steps are followed in problem formulation
then a problem is said to be formulated by an approach
called as state space approach.
• Find Problem solution path in the state space.
• Various problems like planning, learning, theorem
proving, etc, are all essentially search problems.

 Problem Formulation – Vacuum Cleaner

Problem
Definition :
• Consider just two
location of
vacuum cleaner.
• Each location may
or may not
contain dirt and
agent may be in
one location or the
other.
• The goal of the
agent is to clean
100% of dirt from
both the location.

 Problem Formulation – Vacuum Cleaner

Problem Formulation :
• The state, operators, goal test and path cost as below :
1. States : One of the eight states (integer dirt and robot location)
2. Initial state : Dirt in both locations and the vacuum cleaner in
one of them
3. Set of Rules : L (LEFT), R(RIGHT), S(SUCK)
4. Goal Test : No dirt in any location.
5. Path Cost : Let each action cost be 1 unit.

 Problem Formulation – Vacuum Cleaner

• A state space is drawn such that it represents each state perfectly
indicating the presence or absence of dirt in each of the
locations.
 Problem formulation – Eight Tile Puzzle

Problem Definition :
• The eight tile puzzle
consist of a 3 × 3
square frame board
which holds eight
movable tiles
numbered 1 to 8.
• One square is empty,
allowing the adjacent
tiles to be shifted.
• Find a sequence of tile
movements that leads
from a starting
configuration to a goal
configuration.

Problem formulation – Eight Tile Puzzle

Problem Formulation :
• States: It specifies the location of each of the eight tiles and the blank
in one of the nice squares.
• Initial state: Any state can be designated as the initial state.
• Goal: Many goal configurations are possible.
• Set of Rules / Legal moves : They generate legal states that result
from trying the four actions:
‣ Blank moves left
‣ Blank moves right
‣ Blank moves up
‣ Blank moves down
• Path cost: Each step costs 1, so the path cost is the number of steps in
the path.

Problem Formulation : Missionaries & Carnivals

Problem Definition :
• Three missionaries and three cannibals are on one side of
a river that they wish to cross.
3M3C –> River –> 3M3C
• A boat is available that can hold at most two people and
at least one.
• You must never leave a group of missionaries
outnumbered by cannibals on the same bank.
• Find an action sequence that brings everyone safely to
the opposite bank.

 Missionaries and Cannibals

Problem Formulation :
• States : Configuration of missionaries and cannibals and boat on
each side of river.
• Initial State : 3 missionaries, 3 cannibals and the boat are on the
near bank
• Rules : Move boat containing some set of occupants across the
river (in either direction) to the other side.
• Goal state : Move all the missionaries and cannibals across the
river.
• Path Cost : When number of cannibals changes is 6 and number of
missionaries change are 4.
• Constraint: Missionaries can never be out numbered by cannibals
on either side of river, or else the missionaries are killed.

 Design Issues of Search Program

• To choose the most appropriate method for a particular
problem. It is necessary to analyse the problem along the
several key dimensions.
• Some main features of a problem are listed as follows :
1. Is the problem decomposable into set of sub problems?
2. Can the solution step be ignored or undone?
3. Is the problem universally predictable?
4. Is a good solution to the problem obvious without
comparison to all the possible solutions?
5. Is the desired solution a state of world or a path to a state?
6. Is a large number of knowledge absolutely required to solve
the problem?
7. Will the solution of the problem required interaction between
the computer and the person?

 Searching for Solution

▪ So far we have talked about how a problem can be
looked.
▪ Having formulated some problems, we now need to solve
them.
▪ A solution is an action sequence, so search algorithms
work by considering various possible action sequences.
▪ The possible action sequences starting at the initial state
form a search tree with the initial state at the root.
▪ The branches are actions and the nodes correspond to
states in the state space of the problem.

 Searching for Solution

• Find a route from Arad to Bucharest.
• The root node of the tree corresponds to the initial state, expand
through nodes until either a solution is found or there are no more
states to expand.

 Searching for Solution

• Repeated
state
• Loopy
path
• Redundant
path

Searching for Solution

 Searching for Solution
• A sequence of search trees generated by a graph search
• The separation property of GRAPH-SEARCH, The frontier
(white nodes) always separates the explored region of the state
space (black nodes) from the unexplored region (gray nodes).

 Searching for Solution

• Search algorithms require a data structure to keep
track of the search tree that is being constructed.
• For each node n of the tree, we have a structure that
contains State, Parent, Action and Path cost.
• There are many ways to represent nodes. Assume
that the node is the data structure with the following
five components :
• The state in the state space to which the node
corresponds.
• The node in the search tree that generated this node,
this is called as parent node.
• The operator that was applied to generate the node.
• The path cost of the path from initial node to the node.
• Date type : Node
• Components : State, parent node, action, depth, path-cost.

 Searching for Solution

• Queue is used so that search algorithm can easily
choose the next node to expand according to its
preferred strategy.
 Searching for Solution
• Measuring problem-solving performance
‣ Completeness: A search algorithm is complete if it finds a solution
whenever one exists.
‣ Optimality: A search algorithm is optimal if it returns a minimum-
cost path whenever a solution exists
‣ Time complexity: How long does it take to find a solution.
‣ Space complexity: How much memory is needed to perform the
search.
• Complexity is measured by b branching factor (maximum number of
successors of any node), d depth of the shallowest goal node (the number
of steps along the path from the root), m maximum length of any path in
the state space.
• Time is often measured in terms of the number of nodes generated during
the search, and space in terms of the maximum number of nodes stored
in memory.

 Search Strategies
• Searching is the process to find the solution for a given set of
problems. This in AI can be done by using either uninformed
searching strategies or informed searching strategies.
• Uninformed Search / Blind Search :
• This method has no information about the number of steps or path cost
from the current state to goal state.
• Not aware that search process is moving in the direction of the goal or
in the opposite direction.
• It only checks for the state generated is goal or not. If so it will stop
otherwise continues to generate next state.
• Informed Search :
• This method knows whether one goal state is more promising than
other is called a heuristic / informed search techniques.
• They are aware if the search is moving in the right direction or opposite
direction
• It can choose the non goal state that is most closest to the goal as the
next state.
• If the state generated is a goal state, the search process stops else
continues to generate next state in a state space.