Scribd
Upload
8 views

01 CNN

Uploaded by

Jamila Hamdi
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
8 views

01 CNN

Uploaded by

Jamila Hamdi
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Market Basket Analysis & Lift

Start-Tech Academy
Market Basket Analysis & Lift

Start-Tech Academy
Convolutional Layer

Start-Tech Academy
Stride
Stride of 2

Stride of 4

Start-Tech Academy
Padding
Stride of 3
‘Same’ Padding

Stride of 3
‘Valid’ Padding
Padding

Start-Tech Academy
Stride / Padding

Stride:
Stride denotes how many steps we are
moving in each steps in convolution.
By default it is one.

Padding:
Padding is a process of adding zeros to
the input matrix symmetrically to
maintain the dimension of output as in
input

Start-Tech Academy
Filter

? 0.4

0.1

0.2
0.3

0.2

0.3
0.1

0.4

0.3
0.5

0.1

0.4
0.8

0.2

0.4

0.7 0.9 0.8 0.5 0.5

0.9 0.1 0.2 0.1 0.9

Start-Tech Academy
Filter
1 1 1 0 0

0 1 1 1 0 1 0 1 4 3 4

0 0 1 1 1 0 1 0 2 4 3
0 0 1 1 0
1 0 1 2 3 4
0 1 1 0 0
Feature Map

Start-Tech Academy
Feature Map
Feature Map:
• The feature map is the output of
one filter applied to the previous
layer.
• A given filter is drawn across the
entire previous layer, moved one
pixel at a time.
• Each position results in an activation
of the neuron and the output is
collected in the feature map
• A Conv Layer can have more than 1
feature map

Start-Tech Academy
Channels
Channels
• Red, green, and blue can be combined in various
proportions to obtain any color in the visible
spectrum
• Each pixel of any colored image have its own RGB
value that is responsible for the color of that pixel
• For example, the following RGB value might be
used to create purple:
• R: 132 (84 in hexadecimal)
• G: 17 (11 in hexadecimal)
• B: 170 (AA in hexadecimal)

Start-Tech Academy
Pooling Layer
Pooling Layer:
Pooling layers goal is to subsample (i.e.,
shrink) the input image in order to
reduce the computational load, the
memory usage, and the number of
parameters (thereby limiting the risk of
overfitting)

1. Max pooling: The maximum pixel


value of the batch is selected.
2. Average pooling: The average value of
all the pixels in the batch is selected.

Start-Tech Academy
Popular CNN Architectures

1. To build intuition for good model architecture


WHY 2. To use these pre-trained models without retraining for our
problem

Start-Tech Academy
Popular CNN Architectures

The ImageNet Large Scale Visual Recognition Challenge or ILSVRC


for short is an annual competition helped between 2010 and
2017 in which challenge tasks use subsets of the ImageNet
ImageNet Large Scale dataset.
Visual Recognition
The goal of the challenge was to both
Challenge (ILSVRC) 1. Promote the development of better computer vision
techniques
2. To benchmark the state of the art

Start-Tech Academy
Popular CNN Architectures

Popular ILSVRC
submissions

Start-Tech Academy
LeNet-5
Most popular CNN architecture
Created by LeCunn in 1998

Start-Tech Academy
VGG16

Start-Tech Academy
GoogLeNet

Start-Tech Academy
Transfer Learning
Classifier – Fully Connected
Convolutional Base neural network
Convolutional base can be re-used
with new classifier

Advantages
• Saves a lot of training time
• Proven models with good
accuracy
• Models trained on large datasets –
better features extracted
• Easy to use

Start-Tech Academy

You might also like