5 Pro Tips on Working with an AI/Machine Learning Vendor

5 Pro Tips on Working with an AI/Machine Learning Vendor

AUTHOR: Tom Siegman, EVP, Innovation Strategy & Client Relations, RSG Media

At the DPP conference in Dulles, broadcasters discussed how to work with machine learning (ML) vendors. Their critical insight was that ML requires training the model your actual data. Here are 5 “Pro Tips” from the discussion.

Tip #1. Long proofs of concept (POC). Do not expect an out-of-the-box POC. That’s not how ML works.

Tip #2.Realistic Goals. Understand your real upside and set clear benchmarks to know how much to invest in setting up the ML. Ask yourself if you need 100% accuracy or a lower degree will still offer you an advantage. Is that enough if the ML can do 50% of the work for you?

Tip #3. Partner Selection. Choose a great vendor with whom you have a solid relationship. Make sure that each side has skin in the game.

Tip #4. Training Data. Be prepared to supply copious quantities of data to train the model. The more data the model has, the more accurately it can do its work.

Tip #5. Feedback. Part of training a model is offering timely feedback. The model needs to know what it got wrong and what it got right.

RSG Media

RSG Media: The world’s leading media & entertainment companies understand that their success is driven by both art and science. Since 1985 they have trusted RSG Media’s industry experts, data scientists, and technologists to help them drive revenue and increase profits from their rights, audiences, and advertising.

All author posts
Closed

Please Enter Your Email To Access This Content