Jeremy Howard is an entrepreneur, business strategist, developer, and educator. Jeremy is a founding researcher at fast.ai, a research institute dedicated to making deep learning more accessible. He is also a faculty member at the University of San Francisco, and is Chief Scientist at doc.ai and platform.ai.
Previously, Jeremy was the founding CEO Enlitic, which was the first company to apply deep learning to medicine, and was selected as one of the world’s top 50 smartest companies by MIT Tech Review two years running.
An email service offering paid email accounts for individuals and organizations.
A platform for predictive modelling and analytics with over 100,000 data scientists who compete in predictive modelling competitions.
The university's mission is to assemble, educate and inspire a new generation of leaders who strive to understand and utilize exponentially advancing technologies to address humanity’s grand challenges.
A multi-stakeholder community of more than 700 exceptional young leaders who share a commitment to shaping the global future.
Enlitic leverages recent advances in machine learning to make medical diagnostics faster, more accurate, and more accessible.
Jeremy teaches deep learning to students of the Master of Science in Analytics (MSAN) program as well as to the San Francisco data science community more widely.
A research lab working to make deep learning more accessible and widely applicable.
Doc.ai accelerates the transformation of healthcare by leveraging the power of machine intelligence.
platform.ai allows anyone to create world-class deep learning models without code.
Enlitic was the first startup to apply deep learning to medicine. How did you know that deep learning was going to help?
I spent a year figuring that out. I spent a year just doing research. I was totally convinced deep learning was going to be used everywhere. So where's the best place to apply it right now? And specifically, I was looking for places that could have the largest societal impact. Using the technology as it was available at that time, I very quickly realized that it really had to be in vision applications because that was where it had totally proven itself at that time. As a startup, you don't want to do new research to find out whether something's possible.
I always felt like neural nets was the right way to solve the predictive modeling problem. And I definitely felt like we needed to do it on GPUs, but GPUs were so hard to program.
I want to move on to fast.ai. fast.ai has this incredible tagline, "make neural networks uncool again." What was really the vision behind it?
Well, it came out of my frustration with my failures at Enlitic. I had this much bigger vision than radiology. It was really to transform how diagnosis and treatment planning was done throughout medicine. And it was clear I wasn't going to be able to achieve that at Enlitic. As a startup, I couldn't get access to the data that we needed. I found places that had the data and really just came down to, "we're not going to share it with the company because we don't want you profiting from our data." It also was totally incompatible with the incentives of both the investors and the staff who saw we had a product market fit with radiology. "Why do you want to go to something else? Like focus on the thing that's working great."
It's not to hard to learn to become an effective deep learning practitioner. So we have lots of examples, like lots and lots of examples of people doing that through our courses in two months.
Would you recommend that software teams search for specialized sort of machine learning talent? Or should they retrain and teach their existing engineers how to do more deep learning?
Definitely lean towards the latter, which is retrain existing internal people. It's not too hard to learn to become an effective deep learning practitioner. We have lots of examples of people doing that through our courses in two months. They have to be a pretty strong coder already. They have to be tenacious. It's luck. It's not some magic thing, where you don't have to work. But if you've got that, you can be an effective deep learning practitioner in a couple of months. Particularly if it's an an area where the kind of domain is somewhat well explored, like audio image, text, Tabula or collaborative filtering. Something like genomics is going to be a lot tougher because it's not really sorted yet.