This Week in Data — Talent Wars, new chips, and MLConf
My thoughts on what’s timely, interesting or quirky in the world of data. If you’d like to receive this in your inbox every week, subscribe here.
Kick Off
The talent war among start-ups and tech companies increasingly resembles what was going on in finance ten years ago. I remember it well — I spent 2003 to 2008 working for a hedge fund. Here’s a good story in The Economist talking about Silicon Valley fighting for top talent from universities. The concern is that it’s not great for the world if too much talent flocks to one industry. We need bright minds spread around.
In the News
NVIDIA introduced a new chip, the Tesla P100, with 15 billion transistors for deep-learning computing. It’s the biggest chip ever made. Thousands of engineers worked on it for years. Big implications here for Artificial Intelligence.
Here’s a good Forbes interview with a co-founder of Cloudera, which was early to the model of providing services tied to Open Source code. He reflects on how far industry has come since 2008, when the company started, saying: “In the early days we had to be super-evangelical. Why does data matter? Why do we need so much of it, and why is this platform the right approach?”
It’s great to see Intel opening another parallel computing center for research. Industry is ahead when it comes to running analytics on big data sets across lots of computers and it’ll make a huge impact on public research to bring academic computing centers up to speed. This new Intel center will focus on earthquake research in California. A similar partnership to bridge the knowledge gap between industry and academia was introduced this week between University of Michigan and IBM.
In Industry
J.P. Morgan Chase’s annual report has a section that reads more like a tech company’s annual report than a bank’s. That’s because the bank’s CEO, Jamie Dimon, has built out a huge data unit at the bank. Worth reading about the bank’s data strategy.
J.P. Morgan isn’t just being savvy. Banks are actually facing a real competitive threat from tech companies trying to break into finance. Here’s a good New York Times column on this, and The Times had an interesting story on start-up tech firms serving the unbanked.
There are lots of new big data products out there. This site has a good slideshow about a lot of them (warning it is a bit technical). There are so many options available, each of which is good only for certain tasks, that it’s become more important for companies to figure out their data goals before picking the products they’ll use. That’s one of the things we have been advising companies on.
Quirky Corner
As a math major myself, I enjoyed reading about the Mathematical Genealogy Project,which uses data mining to examine the evolution of math going back several centuries.
What’s happening at Ufora
I will be presenting at the Machine Learning Conference in NYC next Friday. If you’d like to attend, you can get 18 percent off by registering with this code: ufora-mlconf. It’s a great conference, and a chance to see what’s going on in the absolute cutting edge of machine learning.
