Video: Best Practices for Training Your AI

Learn more about AI platforms at Streaming Media's next event.

Watch the complete video of this presentation, AI101: From Content Creation to Delivery: How AI Is Impacting Modern Media, in the Streaming Media Conference Video Portal.

Read the complete transcript of this clip:

Cullen Gallagher: There's sort of this push and pull between the business and the technical side. On the business side, you want to be constantly listening to your customers, asking what they want, implementing those things. But that has an effect on the technical side that you have to account for. So for example, we had a hockey customer that said, "Hey, can you pull out all the face-offs?" And we said, "Yeah, sure." So we taught a classifier how to understand just from the video, no metadata at all, when a face-off was occurring, and clipped those out.

That's one of those things when you go and you change the model, it starts to do other things into the neural networks, and in your total package that you have to worry about from a technical standpoint. For us, we ran into things like overfitting of the model, so now it was keying in and seeing in many thousands of different games, but scoreboards, now that we've implemented this new module, it started to hone in only on the kind of scoreboards that it knew, and became inflexible and rigid.

When you're creating your AI strategy and using the best practices and things like that, just understand the push and the pull between the product requirements and the business side stuff, with what it's going to do on the technical side, and how it's going to affect your product in the long run.

Josh Gray: All right, I want to add one thing to that, that overfitting problem is an incredibly common first result when you're going through some of these exercises. And a lot of that is related to the selection of your training dataset relative to your output. If what you're after is looking for scoreboards, and you train a bunch of pictures, they have variety but the scoreboards are always prominent, then when you start throwing real images where maybe the scoreboard is a little more off to the side, or different angles, you might find that you have overfit to a very clean scoreboard classifier. As you're understanding what you want to get out of it, make sure that you're throwing the right dataset into it, to make sure that you get the variety that you want it to be able to handle.

Jason Hofmann: After you've selected the data that you want to train it with, there are also best practices for how to train your algorithm, like, for example, using a random subset of the training data instead of all of the training data, or using five random overlapping subsets of the data to train it, and then test it against some of set of data that it's never seen. So there are best practices that a lot of the workbenches that are out there, point-and-click workbenches. There's quite a few, like from our MATLAB, others, a lot of them will guide you through that, and say, "I know you have a million data points. Don't train me on all of them. Let's do 15 iterations of 10,000 data points each, see what we come up with, and then let's go see how it works against 100,000 data points I've never seen before."

Streaming Covers
Free
for qualified subscribers
Subscribe Now Current Issue Past Issues
Related Articles

Does Facebook's Video AI Suppress Some Clips? BET Says Yes

Network clips that display tune-in information are automatically suppressed by Facebook's AI, says BET, forcing the network to spend more on promotion.

Video: What's the Difference Between Machine Learning and AI?

Microsoft's Andy Beach and IBM/Watson Media's Ethan Dreilinger break down the differences between machine learning and AI in this clip from their panel at Streaming Media West 2018.

Video: How to Know When AI Isn't the Solution to Your Problem

Google's Matthieu Lorrain cautions of the risks of doing AI for its own sake in this clip from Streaming Media West 2018.

Video: Key Considerations When Choosing a Video AI Platform

RealEyes Director of Technology Jun Heider discusses the importance of internal self-assessment and which use-case elements to consider when choosing a platform for video AI in this clip from Streaming Media East 2018.

Video: Who Are the Key Players in Video AI?

RealEyes Media Director of Technology Jun Heider identifies the key players in the AI platform space in this clip from Streaming Media East 2018.

Video: Tips for Getting Started with Video AI Platforms

RealEyes Director of Technology Jun Heider outlines the first steps in choosing an AI platform in this clip from his presentation at Streaming Media East 2018.

Video: How Reinforcement Learning Enables Personalized Viewing Experiences

Microsoft Principal Product Manager Rafah Hosn makes the case for reinforcement learning as a machine learning paradigm for content personalization in this clip from Streaming Media East 2018.

Video: Pros and Cons of Supervised Machine Learning for Content Personalization

Microsoft Principal Product Manager Rafah Hosn discusses the benefits and limitations of a content personalization strategy based on supervised machine learning in this clip from Streaming Media East 2018.

Video: How Microsoft's Custom Decision Service Improves Content and Ad Engagement for Brands

Microsoft Principal Product Manager Rafah Hosn explains how Microsoft's machine learning-driven decision services helps brands target viewers and increase engagement in this clip from Streaming Media East 2018.

Video: How Do We Define Quality of Experience for Streaming Video?

Comcast Technical Solutions Architect Ribal Najjar defines video QoE both in terms of subjective experience and qualitative measurement in this clip from Streaming Media East 2018.

Video: How IRIS.TV Implements Machine Learning in Production Environments

IRIS.TV CEO & Co-Founder breaks down discusses IRIS.TV's approach to helping traditional media companies capture and leverage audience data and machine learning in this clip from Streaming Media East 2018.

Video: How USA Today Leveraged Video AI at the 2018 Winter Olympics

Gannett Senior Director Kara Chiles discusses how USA Today leveraged IRIS.TV and data to localize and personalize their Winter Olympics 2018 coverage in this clip from Streaming Media East 2018.

Video: How to Use Machine Learning to Create Personalized TV Experiences

ZoneTV's Tom Sauer describes how machine learning can be used to overhaul the TV world and deliver more individualized experiences in this clip from Streaming Media East 2018.

Video: How AI Can Open Up New Markets for Your Video

REELY CEO Cullen Gallagher makes the business-growth case for content owners developing an AI strategy in this clip from Streaming Media East 2018.

Video: How IBM is Using Video AI

IBM Watson Media's David Clevinger discusses how media entities are currently using video AI in this clip from Streaming Media East 2018.

Video: How Video AI Helps Businesses Interpret Experience Metrics

Citrix Principal Architect Josh Gray explains how video enables higher-acuity metrics analysis in this clip from Streaming Media East 2018.

Video: How Video AI Improves Content Delivery Efficiency

Limelight VP of Architecture Jason Hofmann discusses how AI impacts content delivery optimization in this clip from Streaming Media East 2018.

Video: Best Practices for Developing Machine Learning Algorithms for Video

Citrix' Josh Gray provides tips on AI model development and Reality Software's Nadine Krefetz and IBM's David Clevinger speculate on the possibilities of metadata-as-a-service in this clip from Streaming Media East 2018.

Video: How Will Machine Learning Impact the Media Supply Chain?

Google's Leonidas Kantothanassis explores the vast range of applications for machine learning in the media workflow and supply change in this clip from his Content Delivery Summit keynote.

Companies and Suppliers Mentioned