👉What is drift?
👉How does non-stationarity fit in?
👉Machine Learning (ML) models can decay?

...these are all valid questions, and, yes, ML models do have to be maintained. Moreover, ML best practices encourage treating models more like living things, continuously monitoring their health, and remediating problems as they arise.

 

Now, that's easy to say...but the near-inevitable degradation of ML models in production is not an easy thing to understand. Why and how does it happen? Remember the old adage, "a picture is worth a thousand words?" Yeah, visuals really help!

 

So, join us for this live presentation where we will visually explain the challenges of ML drift & model decay!

  • Date: Friday, March 18th
  • Time: 2pm ET / 11am PT
  • Length: 1 Hour

Register for the Session:
Ted Hallum

About our Guest Speaker:

Ted Hallum is a Senior Defense Machine Learning (ML) engineer at Octo where his current work focuses on the challenges of data/concept drift detection, model decay, streaming outlier detection, model explainability, active learning, etc. While beyond the conventional ML workflow, these additional due diligence steps are critical to detecting & remediating problematic models. Ted is a former mid-career defense intelligence analyst, a U.S. Army veteran, Founder of the Veterans in Data Science & Machine Learning community, Host of The Data Canteen podcast, and holds a Master of Science in Business Analytics from the College of William & Mary.
AndrewHeadshot

"AI is the new electricity."
Dr. Andrew Ng

About FourthBrain: 

To launch a career in ML you need more than just technical knowledge: you need the real world practical skills that companies are hiring for today. At FourthBrain we give you these skills by taking learning to the next level. Backed by Andrew Ng’s AI Fund, visionaries in the AI industry, you will get the skills you need to succeed. Our unique part-time Machine Learning programs will teach you the technical, business and communication skills that are in high demand today. 

Register for the Session: