Practice makes perfect. It’s true for people learning to master a new skill, and it’s also true for your AI brain. Just as you need the right environment to practice, get feedback and try again, so does your AI brain. In this course, you’ll solve industrial engineering problems inspired by real problems your instructors have worked on in industry.
Class Deals by MOOC List - Click here and see Coursera's Active Discounts, Deals, and Promo Codes.
You’ll learn how to build, test and deploy an AI brain using Microsoft Bonsai, a cloud-based, low-code platform. We’ll walk through the entire Bonsai platform from setup to deployment. Along the way, you’ll use Bonsai to conduct machine teaching experimentation to train a brain and assess its progress. Because you’ll be teaching the brain a relatively complex task, you’ll run multiple simulations until you’re satisfied with the results. You’ll then prep the brain for graduation into the real world — deploying it into a machinery control system or other live environment.
At the end of this course, you’ll be able to:
• Build an autonomous AI that combines reinforcement learning with machine learning, expert rules and other methods that you’ve used in the first two courses of the specialization
• Establish requirements for a simulated environment for your brain to practice a task
• Validate and assess your brain’s performance of a task and make improvements to your brain design
• Evaluate whether a simulator is a good practice environment
• Deploy a brain on a real piece of hardware
This course requires an Azure subscription.
This course is part of a specialization called Autonomous AI for Industry.
Course 3 of 3 in the Autonomous AI for Industry Specialization.
What You Will Learn
You’ll build an industrial strength AI brain using complex features and simulations to solve real-world problems.
Syllabus
WEEK 1
Build Your First Brain
You'll begin this course by implementing a monolithic (one concept) brain.
This week you'll setup your Bonsai workplace and train a brain. By the end of this week you'll be familiar with the Bonsai user interface, and learn how to read the assessment graphs that tell you how the brain is performing.
WEEK 2
Industrial Strength Brains!
This week you're going to train a much more complex brain, something more like the kind of brains that you train for real industrial processes. These brains require decomposition of the task into skills, training or programming multiple skills, and experimentation to get the orchestration and the training right. Juan Vergera and John Alexander will guide you through the steps. You'll practice, and then you'll take a quiz that helps you determine whether you have a good understanding of the steps that you completed so you can do it again on other projects.
WEEK 3
Machine Teaching experimentation
Brain designs are like lesson plans that guide students in their learning. After developing a Lesson plan and teaching it to a variety of students, a teacher will track how well different types of students learned different skills and concepts under different conditions and then modify that Lesson plan for the next set of students.
Machine teachers design AI brains based on subject matter expertise that will guide your students’ learning. This is your original Lesson plan. But then based on how the brain learns, you'll modify that Lesson plan. You'll modify the brain design, so that it can learn even better next time. We call this machine teaching experimentation and that's the primary focus of this week. Learning in autonomous AI brains takes place in layers. There's a teaching layer and a learning layer. The teaching layer sets the structure of which skills the brain needs to learn. That's your brain design. The learning layer is algorithms that learn by practice, neural networks, programming and math. Through this trial and error, the brain is built.
WEEK 4
Simulations: Creating the Classroom for your brain
This week we're going to discuss the ins and outs of simulators. Why do we use simulators for autonomous AI to practice? What kinds of simulators do you need to understand and be familiar with in order to train autonomous AI? At the end of this course you will actually practice connecting a simulator to the project bonsai platform for training.
WEEK 5
Graduating to the Real World
This week you'll practice mining training logs, learning what's working in your training, what's not working in your training, and you're gonna use that information to update your training plans.
Your assessments will get more complex, too, from the automatic assessments in the platform to custom assessments that will analyze specific scenarios, even rare scenarios that are keeping your brain from succeeding well under all conditions.
