Recently, a machine learning algorithm went wild live in front of an audience of millions, as the cartoon South Park trolled Alexa in a really funny way. In the episode the iconic character Cartman kept asking Alexa to add items to his shopping cart. However, viewers’ Alexas near their TVs responded in the same way, adding the items that Cartman ordered to THEIR shopping carts.
The problem with Alexa is that it requires context. Machines lack the context of how the world actually behaves and are very limited in their scope. Yet by monitoring their own machine learning algorithms, Amazon could have easily tracked how Alexa units around the country were adding similar items at a massive scale.
The Alexa issue is one example of a far larger issue. Today, businesses are collecting unprecedented amounts of data. While this data is undeniably useful, the biggest issue is noticing when something strange is happening. Anomalies constantly occur, and detecting them fast is really critical. Ira Cohen explains how to catch problems and glitches early on by using machine learning algorithms to monitor these algorithms for anomalous behavior.
Ira Cohen is a cofounder and chief data scientist at Anodot, where he’s responsible for developing and inventing the company’s real-time multivariate anomaly detection algorithms that work with millions of time series signals. He holds a PhD in machine learning from the University of Illinois at Urbana-Champaign and has over 12 years of industry experience.
©2018, O’Reilly UK Ltd • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • email@example.com