This keynote was presented at the first GE Industrial Machine Learning Workshop on October 24, 2017 in San Francisco, CA during GE's Minds + Machines 2017 Event.
In the field of data science, we are constantly bombarded with innovative approaches and methods. These fresh new tools promise to yield impressively accurate results—and, like all Faustian bargains, can come at a cost that all too often stays hidden until it’s too late. Those of us who gravitate towards this discipline can get easily seduced by the promise of cutting-edge precision. We can even fall prey to our impulses of following the latest and coolest at the expense of our objectives. When does it make sense to deploy complex solutions into production environments? And how should we assess the pros and cons of doing so? The end-goal of this talk is to minimize the chance of adding unneeded complexity to our already constrained systems. To do so, we will review common pitfalls, a few strategies for making complexity assessments, and a framework to make it more likely that the simplest solution possible is implemented to meet every team's objectives.
This presentation was video recorded. You can watch the videos at the following link:
MassTLC Applied Machine Learning Developer Day 2018: https://youtu.be/QfOPS80Jmjw
More topics TBD
- Unsplash: http://unsplash.com
- Wikimedia: http://wikimedia.org
- Fair use of commercial images
And since I'm neither a graphic nor a web designer, I can't take any credit for the design template—only for the contents of my talk. If you like the slides you can get the template here: https://www.goashape.com/shop/kaspian-keynote-presentation-template