The ability of a machine to think and learn is called artificial intelligence.

Machine Learning is an application of AI.

AI is the science and engineering of making intelligent machines.

Usage of Web Data.

Our Browsing pattern is called browsing signature or browsing habit.
Browsing habits, interests and buying preferences is called analytics.

Browser finger printing are:

Cookies
HTML5 Canvas fingerprinting
IP address of the device,

AI word coined by Stanford researcher John McCarthy in 1956.

Why AI Today?
Advanced Tech, better R&D, faster computer, large volume of data.

AI Challenges:
Retain the facts as knowledge.
Recall the knowledge in a situation. Think, analyse and apply logic.
Make useful and accurate predictions.
Make decisions and upgrade their intelligence algorithm themselves.

Sensing: not just about simple input like scanners and sensors. The purpose of sensing is determined by the
Intelligence. Image scanner, audio sensor, speech recognition engine, fingerprint, recognition program,
motion sensor, thermal sensor, light sensors, barometric sensors etc. are the equipment which play central
role where a machine is designed to receive various stimuli from its surroundings.

Perception:
Humans perceive their surroundings with their sensory organs. Then the objects that make the
Surroundings are identified and recognized depending on the retained knowledge about the world. A
machine can have artificial sensory organs like cameras, scanners, photo sensors for light, thermo-sensors
for temperature etc. Think of a robot or machine designed to move in a closed area like office or factory,
more complex environments are railway platforms & airports and most complex of them is a busy road.

Learning:
Humans learn in many ways - guidance and training by others or by self-paced trial and error method. They
retain the learning by practice, remembering and applying it in various situations. Machines are being
developed to learn by trial and error. For example, a machine playing a strategy game like chess may keep
looking for a move that matches the closest correct move and stores it for further usage.

Problem Solving:
In a simple situation, a machine can be programmed into looking for, finding and applying the possible
steps of solution to achieve a set goal. Such machines are useful in a specific task-oriented environment like
bottling plant, loading/ unloading of items, counting assembling parts etc.
Reasoning:

Reasoning
Logical reasoning is the distinct characteristic of human brain- Reasoning has broadly 2 types:
Deductive and Inductive. In deductive reasoning the facts are analysed and guarantee a conclusion.
For ex:
Raj is a non vege and so he will also eat a vegetable if non—vegetarian dish is not available.
All bats are mammals. All mammals give birth to young ones; therefore, all bats give birth to young ones.
Obtuse angles are more than 90 degrees, this angle is 120 degrees so, it must be obtuse angle.

In inductive reasoning. Facts only support the conclusion vvithout any guarantee- For example:
Ravi falls sick most often he eats eggs. Ravi must be allergic to eggs.
Fish is a non—vegetarian dish, Rajesh loves to eat non-vegetarian dishes; therefore, Rajesh
loves to eat fish.

Language:

Learning any language is a complex process and contains grammar and words - word with multiple
meanings, words with similar meanings (synonyms), similar sounding words etc.
Al based voice response systems and chat bots etc. are being developed in a restricted application area but
there is still a lot needs to be done.

Artificial Neural Networks (ANN)

In 1954, at Massachusetts Institute of Technology (MIT), Farley and Clark developed a
128-neurons system which could memorise simple patterns and distinguish one among
many. ANN has paved the foundation for machine learning.

Browsing signature/ browser fingerprinting: Pattern of a user's browsing habit.

Analytics: Analysis of enormous amount of data to produce useful patterns.
Algorithm: A process or logical set of rules to solve a problem or perform a calculation.
Sensing: Perceiving an external stimulus.
Reasoning: Thinking logically to reach a conclusion.
Expert system: A system that compares facts and makes decisions.

AI can be categorised into Complexity of Intelligence and Functionality.

Complexity of Intelligence:
AI with limited functionality - Weaker or narrow AI
AI with human like intelligence - Strong AI

Narrow or Weak AI:

Machines that exhibit a limited extent of intelligence to accomplish a single or simple or repeated tasks or
dedicated assigned tasks.

Some examples are:

• Performing web-wide searchofcontent (e.g. Google search)
• Recognising race among several single image shots or group images,
• Self-driven vehicles.
• Voice interface-based assistants such as Alexa and Siri-
• A robot collecting empty food trays from the table’s restaurant,
• A car driving simulator,
A website suggesting similar products depending on the items bought by the user earlier.

Strong or generalised AI:

Machines with strong intelligence are capable of not only "think" like humans but also able to retain
the learning to solve new problems.
Examples:
A self-Learning diagnostic system loaded with information regarding diseases and possible
Symptoms along with rules to diagnose the diseases,
An airplane training system that functions without the help of a trainer.
An intelligent chat-bot that understands customer's needs and suggests solutions by its
Learned intelligence-
A teaching robot that learns by answering students' queries and thereby enhancing its
teaching skills.

Al Types on the Basis of Functionality:

On the basis of functionality, artificial intelligence can be classified into four types:
 Reactive machines
 Limited memory
 Theory of mind
 Self-awareness.

Reactive machines
As the name suggests the reactive machines knows how to respond to an input on the basis of a set of rules
and the logic. They are most basic type of AI. They do not store learning by previous problem solving. E.g:
Computer play strategy games with humans.

Limited Memory:
Are equipped with the logic of sensing the changes occurring around. Machines can't retain the learning.
Same situation, logic is executed afresh to solve the problem.
e.g. Selfdriven cars, chatbots and customer response systems.
Theory of mind:
Are capable to interacting socially. Able to responds to others by understanding of thier emptions and
gestures.'

Self-awareness:
Are intelligent machines. Donot exits,

Artificial super intelligence - Using which machines will surpass human intelligence due to speed, process
large data and self-evolving smarter algorithms.
IBM Watson: AI based question -answer system based on natural language.
IBM Deep Blue - A chess playing super computer beat Gary Kasparov.

Alexa, Google Assistant, Siri and Cortana are voice driven virtual assistants with speech recognition and
NLP.
Alexa - 90000 functions, Google Assistant - Identifies objects, songs, ecommerce by voice. Siri - navigate
locations.
Cortana - voice based commands and search features.
SELF-DRIVEN CAR -WAYMO: With no driver and any human help.

Humanoids are the robots that look like humans exteriorly. The humanoids closest to humans in
looks are expression are Androids. Humanoids, today, are used for various purposes like teaching, ocean
exploration, carry out search and rescue missions.
How does AI work?
Huge amount of data (e.g. distinguish between cat and dogs)
Filter - Which data is useful?
Algorithms - Identify patterns & trends (statistics and data analytics)

AI and Neurons: Neurons are the microscopic cells that carry info throughout the nervous system and in to
the brain. AI focuses on building a network of artificial neurons.

Machine Learning enables a computer system to learn from experience without programming it further.
Supervised ML - Inputs, what to predict, findings are stored and used for future.
e.g fastest possible route by checking patterns in traffic and road condition ( street, flooded, bad road).
UN Supervised ML - Input, no desired output details, It uses data to find patterns.
Deep Learning machines - New trends and patterns in data. e.g which student are likely to show improved
or declined performance in the next exam.

Reinforcement Learning:
Concept reward punishment-
For each failure is a punishment which helps the machine realize that the step
was wrong while a reward helps machine learn the right step- For example, an AT
system is designed to find the shortest and fastest route between two points.
For every obstacle it encounters, it learns that the movement or turn was wrong
(punishment) so next time it tries the alternate path which lets it move further
distance (reward) until another obstacle. Eventually, the learning of correct
route is reinforced.

Robotics, industrial automation, supply chains, games, information compilation,

automated vehicles are some practical applications of reinforced learning-