Scribd
Upload
26 views49 pages

Convolutional Neural Networks

Uploaded by

hema2110364
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
26 views49 pages

Convolutional Neural Networks

Uploaded by

hema2110364
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 49

Convolutional Neural

Networks
Dr. N. Sripriya
Associate Professor
Department of Information Technology
SSN College of Engineering
[email protected]
Agenda
• Introduction

• Why CNNs are popular?

• Steps to build CNN

• Benefits of CNN

• Research directions
Introduction
Deep Learning is a machine learning technique that can learn useful
representations or features directly from images, text and sound
Why CNN?
Why CNN?
Convolutional Neural Network

• Convolutional neural network (CNN) is a special type of multilayer


feedforward neural network (MLFFNN) that is well suited for pattern
classification.
• A CNN is an MLFFNN designed specifically to recognize 2-dimensional
shapes with a high degree of invariance to translation, scaling, skewing and
other forms of distortion.
Convolutional Neural
Networks
Steps to build CNN
Convolution – Signal processing view
1D Convolution

Input

Filter

output
2D-Convolution Operation
2D - Convolution operation
Convolution operation
Convolution operation
Convolution operation
Convolution operation
Feature Detector/ Filter-1
Feature Detector/ Filter-2
Feature Detector/ Filter-3
Convolution operation
Convolution operation
These are the network
parameters to be learned.

1 -1 -1
1 0 0 0 0 1 -1 1-1 Filter 1
0 1 0 0 1 0 -1 -1 1
0 0 1 1 0 0
1 0 0 0 1 0 -1 1 -1
-1 1 -1 Filter 2
0 1 0 0 1 0
0 0 1 0 1 0 -1 1 -1

…
Each filter detects a
small pattern (3 x 3)
Step 1(B) ReLu layer
• Rectifier is a simple activation function defined as
f (x) = max(0, x)
• Neuron using rectifier is popularly known as ReLU (Rectified Linear Unit)
• It has been shown that ReLU converges faster than networks using
traditional ’tanh’
Step 2 – Max Pooling
Why Pooling?
Why Pooling?

Subsampling
Subsampling pixels will not change the
object
We can subsample the pixels to make image smaller
fewer parameters to characterize the image
Max Pooling
Step 3 - Flattening
Step 4 – Fully connected layer
The whole CNN

Convolution

Max Pooling

Fully Connected A new


Feedforward network image
Convolution

Max Pooling

Flattene A new
d image
Why multiple Convolutional layers?
Learning Multiple levels of representation
Benefits of CNN
• Sparse connections
reduces model parameters and improves generalization
• Locality
Locality in the units of the convolution layer allows more
robustness against non-white noise.
• Weight Sharing
A weight sharing strategy leads to a significant reduction in the
number of parameters that have to be learned.
• Pooling
This leads to minimal differences in the features extracted by the
pooling layer when the input patterns are slightly shifted along the
frequency dimension.
Convolution v.s. Fully Connected
1 0 0 0 0 1 1 -1 -1 -1 1 -1
0 1 0 0 1 0 -1 1 -1 -1 1 -1
0 0 1 1 0 0 -1 -1 1 -1 1 -1

1 0 0 0 1 0
0 1 0 0 1 0
0 0 1 0 1 0 convolution
image

x1
1 0 0 0 0 1
0 1 0 0 1 0
Fully- x2
0 0 1 1 0 0
connected 1 0 0 0 1 0
…
…

…
…
0 1 0 0 1 0
0 0 1 0 1 0
x36
1 -1 -1 Filter 1 1 1
-1 1 -1 2 0
-1 -1 1 3 0
4 0 3
1 0 0 0 0 1
0 1 0 0 1 0 0
0 0 1 1 0 0 8 1
1 0 0 0 1 0 9 0
0 1 0 0 1 0 0: 0
0 0 1 0 1 0
13 0
14 0
fewer Only connect to
15 1 9 inputs, not
parameters! 1 fully connected
1 -1 -1 1 1
-1 1 -1 0
-1 -1 1 0
:4 0 3
1 0 0 0 0 1
0 1 0 0 1 0 0
0 0 1 1 0 0 1
1 0 0 0 1 0 0 -
1
0 1 0 0 1 0 0
0 0 1 0 1 0
0
0
Fewer 5 1
parameters Shared weights
Even fewer 1
parameters
A CNN compresses a fully connected network
in two ways:

• Reducing number of connections


• Shared weights on the edges
• Max pooling further reduces the dimension and
complexity
ImageNet
•About 15 million images belonging to about 22000 categories
•All collected from Web and labeled by human (Amazon’s Mechanical Turk crowd sourcing)
•RGB Images with varying resolutions
•ImageNet Large Scale Visual Recognition Challenge (ILSVRC)
• 1000 categories
• 1.2 million training samples (roughly 1000 per category)
• 50,000 validation images
• 150,000 test images
• Performance Measures: top-1 and top-5 error rates.

Figure : Sample Images from ImageNet


Success of CNN
Research directions using
CNN
Region based CNN
Occlusion study in CNN
Deep Art

Content Style New


Image Image Image
Create a image of a particular
class
Deep Dream
Thank You!!!

You might also like