2. INTRODUCTION
ARTIFICIAL INTELLIGENCE
The art and science of bringing, learning, adaption and self organization
to the machine is the art of artificial intelligence.
Study and creation of computer systems that exhibit some form of
intelligence.
What is intelligence?
Intelligence involves learning, adaption and self organization in
response to known/unknown situations or stimuli.
RATIONALITY
The system is rational if it does the right thing given what it knows
3. APPROACHES OF AI
• THINKING HUMANLY
• THINKING RATIONALLY
• ACTING HUMANLY
• ACTING RATIONALLY
4.
5. ACTING HUMANLY
TURING TEST APPROACH
• The Turing test proposed by Alan Turing in the year of 1950 raised a
question “Can machines think?”
• Turing test is “A computer is programmed well enough to have a
conversation with a human interrogator and passes the test if the
interrogator cannot discern if there is a computer or human at the
other end”
6. To pass the Turing test the computer needs two important techniques:
ROBOTICS
“A robot is a reprogrammable, multifunctional, manipular that is designed to
move materials, parts, tools or specialized devices through various
programmed motions for the performance of a variety of tasks”.
Some form of intelligence is embedded into it
VISION AND SPEECH PROCESSING
Computer vision is gathering meaning full information from images and
videos.
Eg. In health care domain and military.
It involves classical AI methods of symbolic processing.
Eg. Alexa and Apple Siri.
7. CAPABILITIES NEEDED BY THE COMPUTER
NATURAL LANGUAGE PROCESSING
Deals with the methods of communicating with a computer in one’s
own natural language
KNOWLEDGE REPRESENTATION
To store what it knows
AUTOMATED REASONING
To answer questions to draw new conclusions
MACHINE LEARNING
To adapt to new circumstances and to detect and explore patterns
8. THINKING HUMANLY
COGNITIVE MODELLING APPROACH
Determining how the human brain actually thinks
Finding the theory and applying it as a program to the computer.
COGNITIVE SCIENCE
Cognitive science is the combination of computer models from AI and
the experimental techniques from phycology to construct precise and
testable theories of human mind.
9. THINKING RATIONALLY
• INTELLIGENT AGENT
An intelligent agent is a software entity which senses its environment
and then carries out some set of operations on behalf of a user.
It employs some knowledge or representation of the end users goal
Agent is not same as a program
An Agent perceives its environment through sensors. It then acts upon
the environment through actuators
11. CHARACTERISTICS OF AGENT
AUTONOMOUS
Can work on their own.
PERSISTANT
They are persistent over a prolonged period of time.
ADAPTIVE
Adjust to the changes happening around.
• Examples of agents
A human agent has eyes, ears, nose etc. sensors are the hands, legs
and mouth are the actuators.
12. • MOBILE
They can be transported over the networks
LEARN
They have a good ability to learn
AGENT PROGRAM
• Developing the agent program and tabulating the agent functions
• This leads to infinite functions so we need to put a limit to percept
sequence.
• Table of functions of percept sequences and actions are external
characteristics and internally agent function will implement the agent
program.
13. AGENT PROGRAM
Function is used to
store the percepts
in tables
Action is
looking up the
percepts in
the table
14. ARCHITECTURE OF AGENT
• The agent program runs on a computing device which is called the
architecture.
• The chosen program must match the architecture that it will accept
and run
• The architecture helps in making the percepts from the sensor
available to the program, runs the program
• Feeds the program action choices to the effector as they are
generated.
Agent= Architecture+ Program
17. ENVIRONMENTS
• Deals with the environment the agent works.
• A task environment is the problem to which the agent is the solution.
Types of task Environments
FULLY OBSERVABLE VS PARTIALLY OBSERVABLE
Agents sensors give complete access to the environment at every point
of time it is considered to be fully observable
In few environments there is some noise or some inaccurate sensors
Some states of environment are missing then such environment is
partially observable environment
18. Puzzle game environment agent can see all the aspects surrounding
them. Agent can see all the squares of the puzzle game along with the
values
• Eg Image analysis, Tic-tac toe
DETERMENISTIC VS STOCHASTIC
From the current state of environment and action the agent can
deduce the next state of environment
Eg. Image analysis-Current part to remaining part
Stochastic: Not based on the current state
Eg. Boat driving agent-Next driving not based on the current state
The agent takes actions from previous percepts.
19. EPISODIC VS SEQUENTIAL
Episodic environment agent’s experience divided to atomic episodes
Episode has the agent perceiving process and then performing single
action.
Eg. Consider an agent finding the defective parts in assembled
computer machine which does not depend on pervious decisions.
STATIC VS DYNAMIC
Static environments can be tackled easily because there are no big
changes in the environment.
Eg. Static-8 queen puzzle game environment has values in squares
Needs action of an agent
20. • Dynamic-Environment keeps changing-agent does not take any
action.
Eg. Car Driving Agent
DISCRETE VS CONTINOUS
In discrete environment the environment has finite discrete states over
the time and each state has associated precepts and action
Eg. Crossword puzzle.
State changes are continuous
21. SINGLE AGENT VS MULTIPLE AGENTS
• Well defined single agent who takes the decision and acts accordingly.
• Multiagent environment has various agents or group of agents.
• They work in a competitive multiagent environment where agents
work parallelly to maximize the performance
• Co-operative multiagent environment in which all the agents have a
single goal and they work together to get high performance
Eg War games, maze game, fantasy football.
22. APPLICATIONS OF AI
• GAME PLAYING
• SPAM FIGHTING
• LOGISTICS PLANNING
• ROBOTICS
• ROBOTIC VEHICLES
• SPEECH RECOGNITION
• AUTONOMOUS PLANNING AND SCHEDULING
• LANGUAGE UNDERSTANDING AND PROBLEM SOLVING
• DIAGNOSIS
23. TYPES OF AGENT
• SIMPLE REFLEX AGENT
• MODEL BASED REFLEX AGENT
• GOAL BASED AGENT
• UTILITY BASED AGENT
31. PROBLEM SOLVING APPROACHES
• PROBLEM SOLVING AGENT
GOAL FORMULATION
PROBLEM FORMULATION
SEARCH
SEARCH ALGORITHM
OPEN-LOOP
32. • INITIAL STATE
• ACTIONS
• TRANSITION MODEL SUCCESSOR
• SUCCESSOR
• STATE SPACE
• GRAPH
• PATH
• GOAL TEST
• PATH COST
• STEP COST
• OPTIMAL SOLUTION