Convolutional Neural Networks with Keras

Introducing ....

 - Open Source, High-Level deep-learning library in python

 - Sits on top of your normal deep-learning framework                                  (tensorflow, theano, MXNet, CNTK etc)

Why Keras?

 - Allows you to build most model architectures very quickly

 - Removes the tedium of lower-level coding    e.g.

    - No handling of tensorflow session

    - No keeping track of matrix sizes

    - No management of weights and biases

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Before we dive in...

Great documentation at keras.io

Since 2017 keras is integrated as

part of tensorflow core

Large user base and good online support

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Coding in Keras

Raw Tensorflow 2 convolutional layers

And now with Keras

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Coding in Keras

Raw Tensorflow  - Adding a Dense Layer

And now with Keras

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Coding in Keras

Raw Tensorflow  - Loss and Optimizer

And now with Keras

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Coding in Keras

Raw Tensorflow  - Training

And now with Keras

Note on "epochs":  

An epoch (in keras case) implies a run through the entire dataset (55,000 images).

In tensorflow we made 20,000 iterations over batches of size 50.  This equates to 1,000,000 training samples.   

In keras to use 1,000,000 training samples we will need to run through the entire dataset (1,000,000 / 55.000) times which is ~= 18 times (18 epochs)

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Coding in Keras

Raw Tensorflow  - Testing

And now with Keras

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Coding in Keras

Equivalent code in keras

Code to build, train and print accuracy of our network

(tensorflow)

Keelin Murphy

Courtesy Keelin Murphy

@MurphyKeelin

Keras

By keelinm

Keras

  • 109
Loading comments...

More from keelinm