Book Image

Automated Machine Learning with AutoKeras

By : Luis Sobrecueva
Book Image

Automated Machine Learning with AutoKeras

By: Luis Sobrecueva

Overview of this book

AutoKeras is an AutoML open-source software library that provides easy access to deep learning models. If you are looking to build deep learning model architectures and perform parameter tuning automatically using AutoKeras, then this book is for you. This book teaches you how to develop and use state-of-the-art AI algorithms in your projects. It begins with a high-level introduction to automated machine learning, explaining all the concepts required to get started with this machine learning approach. You will then learn how to use AutoKeras for image and text classification and regression. As you make progress, you'll discover how to use AutoKeras to perform sentiment analysis on documents. This book will also show you how to implement a custom model for topic classification with AutoKeras. Toward the end, you will explore advanced concepts of AutoKeras such as working with multi-modal data and multi-task, customizing the model with AutoModel, and visualizing experiment results using AutoKeras Extensions. By the end of this machine learning book, you will be able to confidently use AutoKeras to design your own custom machine learning models in your company.
Table of Contents (15 chapters)
1
Section 1: AutoML Fundamentals
5
Section 2: AutoKeras in Practice
11
Section 3: Advanced AutoKeras

Creating a CIFAR-10 image classifier

The model we are going to create will classify images from a dataset called Canadian Institute for Advanced Research, 10 classes (CIFAR-10). It contains 60,000 32x32 red, green, blue (RGB) colored images, classified into 10 different classes. It is a collection of images that is commonly used to train ML and computer vision algorithms.

Here are the classes in the dataset:

  • airplane
  • automobile
  • bird
  • cat
  • deer
  • dog
  • frog
  • horse
  • ship
  • truck

In the next screenshot, you can see some random image samples found in the CIFAR-10 dataset:

Figure 4.6 – CIFAR-10 image samples

This a problem considered already solved. It is relatively easy to achieve a classification accuracy close to 80%. For better performance, we must use deep learning CNNs with which a classification precision greater than 90% can be achieved in the test dataset. Let's see how to implement it with...