Hands-On AI Part 1: Create Applications with Powerful AI Capabilities

Published: 07/20/2017  

Last Updated: 07/27/2017

A Tutorial Series for Software Developers, Data Scientists, and Data Center Managers

Headline-grabbing advances in artificial intelligence (AI), machine learning, and deep learning are transforming software as we know it. Established technology giants and fledgling startups alike are applying AI in new ways, such as self-driving cars, virtual personal assistants, discovery of new medications, or predicting financial market trends. The list of applications is long and varied, yet it barely scratches the surface of what’s to come.

In this tutorial series, experts in AI, machine learning, and deep learning familiarize you with AI tools, infrastructure, and techniques by demonstrating the process of building an application with powerful AI capabilities.

AI in Action

Using an automated movie-making app as an example, you will learn about two challenging and fundamental AI problems: image classification and sequence prediction. Discover how to use convolution neural networks for image classification by taking a set of images as input and automatically detecting emotions in those images. Next, you will learn to use recurrent neural networks to algorithmically synthesize a musical melody to accompany the emotions in the images. The final output is a finished movie complete with a computer-generated soundtrack.

Diagram of AI music to emotion translation

Throughout the series, we illustrate various AI concepts and introduce you to Intel® architecture that supports deep neural networks. We show you how to help streamline the AI app-coding process using modern technologies from Intel including:

  • Intel® Data Analytics Acceleration Library (Intel® DAAL)
  • Intel® Math Kernel Library (Intel® MKL)
  • Intel® Distribution for Caffe*
  • TensorFlow* deep learning framework

We review the key phases in the AI development process: ideation, team formation, data collection and storage, model development and evaluation, and deployment. We also evaluate all of the critical decision points by comparing various AI algorithms and frameworks, as well as data center and cloud infrastructure options, just as you would while working on your own app.

Required Skills

These tutorials assume an intermediate knowledge of the Python* programming language, basic linear algebra, basic statistics, and basic probability theory, as well as a familiarity with GitHub*. Even if you do not have these skills, you can sign up and follow along as we share source code and environments suitable for cloning to build AI apps of your own. Non-technical readers can gain insights and details on how to develop an AI application. We even introduce you to Docker* containers, the Keras* neural network API, the TensorFlow* software library for machine intelligence, and the Caffe* deep learning framework.

Target Audience

Everyone is encouraged to use these tutorials, but they were written primarily for the following professionals:

  • Data scientists who want to familiarize themselves with Intel’s deep learning and AI software stack and, in the process, see AI technologies from Intel in real applications
  • Software developers who want to get up to speed quickly on machine learning and deep learning by working through hands-on tutorials
  • Product or data center managers who want to experience the new state-of-the-art AI technologies from Intel and learn from experts how to structure an AI project to increase its chances for success

Series Overview

The complete Hands-On AI tutorial series is divided into five phases. Each phase will address multiple focus areas:

  • Idea: Kickoff your project by defining a common set of goals, data sources, technology limitations, the team, and a strategic plan.
  • Technology: Explore various frameworks, learn to select the right one for your project needs and long-term goals, and discover the available data center or cloud computing options for handling your computational needs.
  • Data: Discover how data is collected, analyzed, preprocessed, and annotated.
  • Model: Learn how to set up an experimental infrastructure, pick a model, train it, and iterate until you are satisfied.
  • App: Deploy the model in production with an easy-to-use web interface (for user-facing intelligent apps).
Next: Ideation


View All Tutorials ›

Product and Performance Information


Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.