Book Image

Machine Learning with TensorFlow 1.x

By : Quan Hua, Saif Ahmed, Shams Ul Azeem
Book Image

Machine Learning with TensorFlow 1.x

By: Quan Hua, Saif Ahmed, Shams Ul Azeem

Overview of this book

Google's TensorFlow is a game changer in the world of machine learning. It has made machine learning faster, simpler, and more accessible than ever before. This book will teach you how to easily get started with machine learning using the power of Python and TensorFlow 1.x. Firstly, you’ll cover the basic installation procedure and explore the capabilities of TensorFlow 1.x. This is followed by training and running the first classifier, and coverage of the unique features of the library including data ?ow graphs, training, and the visualization of performance with TensorBoard—all within an example-rich context using problems from multiple industries. You’ll be able to further explore text and image analysis, and be introduced to CNN models and their setup in TensorFlow 1.x. Next, you’ll implement a complete real-life production system from training to serving a deep learning model. As you advance you’ll learn about Amazon Web Services (AWS) and create a deep neural network to solve a video action recognition problem. Lastly, you’ll convert the Caffe model to TensorFlow and be introduced to the high-level TensorFlow library, TensorFlow-Slim. By the end of this book, you will be geared up to take on any challenges of implementing TensorFlow 1.x in your machine learning environment.
Table of Contents (13 chapters)
Free Chapter
1
Getting Started with TensorFlow

Loading a pre-trained model to speed up the training

In this section, let's focus on loading the pre-trained model in TensorFlow. We will use the VGG-16 model proposed by K. Simonyan and A. Zisserman from the University of Oxford.

VGG-16 is a very deep neural network with lots of convolution layers followed by max-pooling and fully connected layers. In the ImageNet challenge, the top-5 classification error of the VGG-16 model on the validation set of 1,000 image classes is 8.1% in a single-scale approach:

First, create a file named nets.py in the project directory. The following code defines the graph for the VGG-16 model:

    import tensorflow as tf 
    import numpy as np 
 
 
    def inference(images): 
    with tf.name_scope("preprocess"): 
        mean = tf.constant([123.68, 116.779, 103.939],  
    dtype=tf.float32, shape=[1, 1, 1, 3], name='img_mean...