Deep_learning_Teacher_ppt.pdf - INTRODUCTION TO DEEP...

INTRODUCTION TO DEEP
LEARNING BASICS
(TRANSITION FROM MACHINE LEARNING TO DEEP LEARNING)

TABLE OF CONTENTS
¡
Deep Learning: What and Why
¡
Key differences between ML and DL
¡
Basic Terminologies
¡
Perceptron
¡
Various Deep Learning Networks
¡
Transfer Learning
¡
Applications of Deep Learning

DEEP LEARNING
¡
Subfield of Machine Learning
¡
Concerned with algorithms inspired by the structure and function of the brain called
artificial neural networks
¡
Attempts to draw similar conclusions as humans would by continuously analyzing
data with a given logical structure

DL ⊂ ML ⊂ ࠵?࠵?

DEEP LEARNING OVER MACHINE LEARNING
¡
Conventional ML methods : succumb to environmental changes
¡
Deep learning : Adapts to changes by constant feedback and improve the model
¡
Deep Learning can be used to solve any complex problem

BASIC TERMINOLOGIES
¡
Neuron:

BASIC TERMINOLOGIES (CONTD.)
¡
Weights:
When input enters the neuron, it is multiplied by a weight
Initialization of weights is done randomly
Updated during the training process
More weight: More important feature
¡
Bias:
Constant added to the NN to change the range of
multiplied input
¡
Activation Function :
Used to “fire” a neuron
Like a gate that checks that an incoming value is greater
than a critical number
Note: A neural network without an activation function is essentially just a linear regression model.