Michael Schmidt: Fitting data with complex machine learning models is solved
Michael Schmidt · Michael Schmidt, SiSense // Data Driven NYC 24 // February 2014 (Hosted by FirstMark Capital) · Mar 3, 2014 · at 9:47
Michael Schmidt, CEO of Nutonian, presents at Data Driven NYC on symbolic regression and the tradeoff between model complexity and interpretability.
0:00 / 0:19exact quote · 19.9s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“So, one interesting thing is it's actually really, really easy to fit data. It's actually, it's quite boring. You know, you have, you know, a really large, complex model. Maybe it's a neural network. Maybe it's a random forest. Or maybe it's just a really large polynomial. So you could have a really complex formula Way out here that captures all the dynamics. It fits it almost perfectly. Like, that's a solved problem.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Michael Schmidt
Nutonian's Eureqa software is installed on 50,000 systems across 90 countries
“So we've been installed over, I think we're up to about 50,000 systems now in about 90 countries, and we're growing by about, you know, 1500 new users, new trials of the software per month.”
Nutonian's Eureqa software is cited in 80,000 independent peer-reviewed publications
“Seeing all the publications that are coming off where they're citing us in the results, so up to 80,000 independent peer-reviewed, you know, publications where we're directly citing the results of this cutting edge research.”
Michael Schmidt's graduate research on symbolic regression was published in Science
“We got published in Science.”
Nutonian helped the U.S. Air Force detect material cracks 10x faster
“One of our other biggest customers is the Air Force, and this is a case where they do a ton of work trying to detect when failures happen, and there's a lot of things that go into failures, and actually being able to detect when cracks form, microscopic cracks…”