The Model Wasn’t Meant For You: Model Accuracy vs Model Interpretability

An incredible demonstration of how humans could potentially interact with artificial intelligence (AI), and what role each plays in a world where they have to co-exist (or destroy one another) can be observed in the HBO show Westworld.

Westworld’s AI is futuristic, but its wisdom about how laypeople understand and interact with sophisticated machine learning is relevant right now.

In this talk, I’ll argue that professional data scientists can learn a few key lessons from Westworld: (1) the models you build for machines may not work for humans; (2) you can and should build “metrics for humans,” too; and (3) you must strive to avoid the uncanny valley—machine learning models that try to blend the two.

** Make sure to register for this free event on Eventbrite: https://www.eventbrite.com/e/the-model-wasnt-meant-for-you-model-accuracy-vs-model-interpretability-tickets-50144031147 **

Meet your Speaker

Victor Amin - Senior Data Scientist - SendGrid

Victor is a Senior Data Scientist at SendGrid, the world's first cloud-based email platform. SendGrid delivers around 2 billion emails per day and generates petabytes of data each year. At SendGrid, Victor specializes in building predictive machine learning models that help business leaders make better decisions.

After graduating cum laude from Princeton University, Victor founded an Internet security company, and later went on to publish on statistical genetics as a bioinformatician at the University of Florida. Victor holds a PhD in Physical Chemistry from Northwestern University, where he studied quantum confinement and applications of machine learning to small molecule discovery.