Download as pdf or txt
Download as pdf or txt
You are on page 1of 49

ARTILLIGENCE

IEEE EUREKA
APPLICATIONS OF MACHINE LEARNING
FACE VERIFICATION
VIRTUAL ASSISTANTS
SELF DRIVING CARS
GOOGLE MAPS
RECOMMENDATION SYSTEMS
WHAT IS ARTIFICIAL INTELLIGENCE

● Artificial intelligence (AI) is intelligence


demonstrated by machines, as opposed
to natural intelligence displayed by
animals including humans.
● Processes and algorithms that are able
to simulate human intelligence.
● Human intelligence "can be so precisely
described that a machine can be made
to simulate it"
WHAT IS MACHINE LEARNING
● Subset of AI that falls within the "limited memory" category.
● Field of study that gives computers the capability to learn without being
explicitly programmed.
● Algorithm improves automatically through experience and by the use of
data
● Build a model based on sample data, known as training data, in order
to make predictions or decisions without being explicitly programmed
to do so.
WHEN TO USE MACHINE LEARNING

1) Problems where there are no human experts, so data cannot be labelled or


categorised.
2) Problems where there are human experts, but it is very hard to program.
3) Problems of a large scale.

Use machine learning to write a program that has to adapt to a fluctuating


environment
REGRESSION
● Determines relationship between a dependent variable and some
independent variables.
● Types:
○ Simple Linear Regression
○ Multiple Linear Regression
○ Polynomial Regression
○ Support vector Regression
SIMPLE LINEAR REGRESSION
● One independent and one dependent variable
● Finds a linear function that predicts the dependent variable as a function of
the independent variable
● Y = aX + b (called best fitting line)
● Least square method used to find the coefficients in the above equation
MULTIPLE LINEAR REGRESSION
● One dependent variable and multiple independent variables
● Y = a1X1 + a2X2 + a3X3 …. + anXn + b
● Y is dependent variable
● X1, X2, X3 … , Xn are independent variables
POLYNOMIAL REGRESSION
● Special case of multiple Linear regression
● Y = a1X1 + a2X22 + a3X33 …. + anXnn + b
● Generally used to predict epidemic spread
● Used if the Linear regression is not able to fit the data properly.
DEEP LEARNING (DNN)
Goal?
How do we get the parameters/knobs?
CONVOLUTIONAL NEURAL NETWORK (CNN)
Let’s take a normal image of 1000x1000x3 pixels and make it as input for a neural network
with two hidden units of 1000 units each. We will have to train 3,00,10,03,001 parameters.

Lots of parameters right?


Can some of these parameters be shared?
Convolution Neural Networks solve this problem
What is Convolution?
Let’s take an example of vertical edge detection
Number of channels for filter == Number of channels for image
REINFORCEMENT LEARNING
RL : EXPERIMENT BY OpenAI
RL : DEFINITION
Reinforcement learning (RL)

is an area of machine learning that focuses


on how you, or how something, might act
in an environment in order to maximize
some given reward.
RL : MARKOV DECISION PROCESSES

Markov decision processes


give us a way to formalize sequential decision making.

Components of an MDP:
● Agent
● Environment
● State
● Action
● Reward
RL : EXPECTED RETURN AND POLICY

Expected Return = sum of future rewards

It is the agent's goal to maximize the expected return of


rewards.
NATURAL LANGUAGE PROCESSING (NLP)

Sentence Classification, Text Classification, Sentiment Analysis, Text


Summarization, Machine Translation and Answer Relations.
Hey why don’t we try a simple Neural net out?

Sentiment Analysis

● Paul is not a good artist.


● Paul is a very good artist.
● Paul is not a very good artist.
RECURRENT NEURAL NETWORKS (RNN)
One-way Recurrent Neural Network
Bidirectional Recurrent Neural Network

Name Identification

● Teddy Roosevelt was a good president


● Teddy bears are on sale
ROAD MAP
ROAD MAP - LINKS
● Linear Algebra: https://lnkd.in/epf9HHWy
● Intro to Machine Learning by Andrew Ng: https://shorturl.at/nLT56
● Neural Networks by 3Blue1Brown: https://shorturl.at/ovzBG
● Deep Learning by New York University (NYU): https://lnkd.in/eZFGQuGZ
● Natural Language Processing by CompSci Oxford x DeepMind:
https://lnkd.in/ehSpESC2
● Convolutional Neural Networks: https://shorturl.at/deqwP
● Reinforcement Learning: https://shorturl.at/quAZ8

You might also like