Wise Insights

Embedded Machine Learning Demonstration at the O'Reilly Strata Conference

Today, wise.io™ is demonstrating our fast machine learning technology on an embedded low-power hardware device. The demo shows off just how fast, portable, and usable our foundational technology is and we hope everyone at the Strata Conference comes by for a viewing starting at 5pm (Tuesday). It’s also a chance for everyone to see a $35 Raspberry Pi® ARM-based device in action, if you haven’t already. Cheeky PR statement: The Internet of Things just got a brain.

Henrik Brink, CTO, said of the demo: “At the same or better accuracy, learning and prediction speeds (and accuracy) with wise.io™ tech on this little 700 MHz ARM device beats out popular ML implementations running on a modern multicore laptop.” Damian Eads, Director of Engineering, said “This is the first time we have shown off one of our more stealthy components of the company and we’re very excited to start getting feedback.”

Embedded Machine Intelligence

There are so many possible directions now that the pandora’s box is open. We’ll give you a little taste here of where we see things going.

rpi.screen

High-speed Parallel Learning and Prediction

If you need to continuously learn on incoming data across a company’s infrastructure, why not simple attach a small low-power ML device to continuously learn and predict? Imagine finding anomalies in router traffic by inspecting every packet. Imagine adding ML prediction device to every node in a large cluster without modifying the framework code.

A Brain for the Internet of Things

The Internet of Things promises interconnected sensors in everything from Smart Cities to toasters. But sending raw streaming data from a vast array of sensors to a central aggregating computer or cluster makes little sense when the questions you want to ask at the device level are already well posed. It makes even more sense when you want to know the answer now, not in retrospect. For example:

  • “Is this accelerometer telling me there’s an earthquake starting?”
  • “Where are the faces in the streaming video I’m shooting?”
  • “Is the purchase on this mobile phone fraudulent given the previous activity and sensor data on the phone?”

We think that adding machine intelligence where the data is generates is the manifest destiny for the internet of things.

Background

wise.io™ is a machine learning company based in Berkeley, California. The startup offers best-in-class machine-learning solutions for enterprise and Machine Learning as a Service™ for data scientists and developers. We’re a finalist at the Strata startup showcase, which kicks off this evening.

RaspberryPi® is a registered trademark of The Raspberry Pi Foundation, a UK registered charity.