Introduction

Deep learning is a subset of machine learning that uses algorithms based on the structure and functions of the human brain. These algorithms are called artificial neural networks, and they learn from large amounts of data. Artificial neural networks include several deep layers that enable them to solve problems that require human-like thought processes.

Deep-learning techniques have the following advantages:

  • Accuracy: Deep neural networks produce highly accurate results in some of the most complex situations, such as image classification.
  • Scalability: Deep-learning methods work efficiently on large volumes of data and are scalable based on the dataset.
  • Robustness: Artificial neural networks can work directly on a raw dataset without requiring any cleanup or engineering.
  • Adaptability: Pretrained deep neural networks can be easily transferred and used with varied domains.

Note

This module's labs can be completed for free using the Databricks 14-day trial, but you cannot use an Azure free trial subscription to create a Databricks workspace. To switch a free trial subscription to pay-as-you-go, go to your profile and change your subscription offer to pay-as-you-go. You may also need to remove the spending limit, and request a quota increase for vCPUs in your region. When you create your Azure Databricks workspace, you can select the Trial (Premium - 14-Days Free DBUs) pricing tier to give the workspace access to free Premium Azure Databricks DBUs for 14 days.

Learning objectives

In this module, you'll:

  • Identify how artificial neural networks work and learn about the use cases for deep-learning algorithms.
  • Train and evaluate a classifier built by using TensorFlow.
  • Create an image classifier by using deep learning.