Computer Vision News - March 2017

32 Computer Vision News RE•WORK Summits in S. Francisco Events Machine Intelligence - Machine Intelligence in Autonomous Vehicles RE•WORK is holding two important summits in S. Francisco on March 23-24 : as usual, you will find the registration links in our Computer Events page . But this time, we decided to tell you more about them. Thanks to the very kind help of RE•WORK’s founder Nikita Johnson , we were able to interview in advance four of the summits’ speakers: Erik Schmidt - Senior Scientist at Pandora, Gilles Backhus - Machine Learning Engineer at Konux and Stacey Svetlichnaya - Software Development Engineer at Flickr will speak at the Machine Intelligence Summit ; Nikhil George - Sr. Manager Perception & Machine Learning at Volkswagen Electronics Research Lab will speak at the Machine Intelligence in Autonomous Vehicles Summit . Read what they say about Machine Intelligence and its future! RE•WORK named the technology behind these two summits “Machine Intelligence”, that is where machine learning meets artificial intelligence. How do you see this meeting between machine learning and artificial intelligence taking place? Nikhil George: For well over a decade, we have been getting more and more reliant on artificial intelligence software - think Google Search for example. However, of late the transition has been happening to machines aka devices where this intelligence is getting more and more “embodied”. We first had Siri - where phones started getting personalities. Then there was Amazon Alexa - where these “personalities” got their very own “body”. The advent of autonomous vehicles will allow these embodied intelligences to now act in our world. I see this transition happening in multiple facets of our world. Gilles Backhus: While "machine learning" (ML) and "artificial intelligence" (AI) are sometimes used as synonyms, I'd rather say that a system / an algorithm / a model designed with ML as a tool could potentially achieve a level of performance / smartness that surpasses human capabilities and thus can be called artificially intelligent. An example would be designing a machine control model (e.g. for a car, plane, CNC machine) that tries to achieve a problem so complex that a human would never be able to find the "formula" themself. However, using supervised machine learning it can very well become possible to find the model that succeeds in its task at hand. In this example, super-human- performance was achieved but in my humble opinion it is still an exaggeration to call it intelligent. True intelligence requires certain traits such as self-awareness and the ability to adapt at all times. It will still take decades until true AI can be achieved. Nikhil George Gilles Backhus Stacey Svetlichnaya Erik Schmidt