Machine Learning Intro

Oh Yes, Machine Learning is here

Yes, A big Yes

Neural Network

But all more simpler

(simple neural network with 2 input and 5 hidden layers)

Input Layer

Each pixel in the image will be a feature (0 ~ 255)

Each image has three channels (R,G,B)

Each channel has 50*50 pixels

Total feature vector size = 50*50*3 = 7500

Hidden Layer

#of layers and # of neurons in each layer are hyperparameters

Need activate function to learn non-linear boundries

Activate Function

Choices for f : Sigmoid, Tanh,

Rectified Linear Unit (f(x) = max(0,x)) ...

The research work

Output Layer

Basically a logistic regression for multiple classes

Inputs are the hidden neurons of previous layers

Cost function is based on the output layer

Output 9 probabilities corresponding to 9 classes

Use back-propagation to calculate gradient and then update weights

Convolution

Can be viewed as a feature extractor

inputs: 5x5 image and 3x3 filter

outputs: (5-3+1)x(5-3+1) = a 3x3 feature map

CovNet Example

Convolution

inputs: 3x50x50 image and k filters with size 5x5

outputs = ???

We will have 3*(50 - 5 + 1)^2*k many features

Need pooling to reduce the # of features

6,348*k

if k = 25, convolution generates around 160,000 features! 

Max-Pooling

Also known as Sub-Sampling

For 2 x 2 pooling, suppose input = 4 x 4

Output = 2 x 2

Architecture

2 convolution layers with pooling

1 hidden layer

1 output layer (logistic regression)

Visualize CNN

Input

First Covolution Layer with 25 filters

Second Covolution Layer with 50 filters

Technologies used

• Python (and its Libraries)
• Tensorflow
• Inception
• Numpy
• Pandas
• and much more.....

Introduction to Inception and Neural network

Inception: The lifeline of Google

Neural Network: The life line of Machine Learning

***In our case we used a Convolutional Neural networks

Inception and Neural network

Working of Inception

A neural network from Scratch

• Installing dependencies
• Merging out Dataset
• Adding Dataset to Inception
• Pulling the changes of Tensorflow

And this is how it looked

The Training and Testing Part

An Accuracy of 88%

Holla !!!!

Result?

Some Graph and analysis

Lets Try it out..

Here is a screenshot of our model classifying a Circle with an accuracy of 99.999%

Hyperparameters

What is a Hyperparameters?

Semi-Supervised

• Unsupervised Pre-training

• Supervised Fine-tuning

1. Use large amount of unlabeled data
2. Work as initialization process of weight

Adjust the weight by cost of classification and back-propagation