Day 1 at the Embedded Vision Summit 2016

May 02, 2016

Max The Magnificent-May 02, 2016

As I pen these words, I'm attending the Embedded Vision Summit at the Santa Clara Convention Center in Santa Clara, California. It's been an incredibly interesting experience thus far, and we haven’t even reached lunch yet.

Following registration, I was ambling around looking for coffee (and hoping for bacon) when I ran into Steve Ernst, who is the CEO of Event Presence. The interesting thing is Steve wasn't attending the conference as a corporeal persona; instead he was using a Beam robot avatar to "Beam In" to the conference.

Yours truly (in the Hawaiian shirt) and Steve Ernst (in the telepresence robot)
(Source: Max Maxfield /

This was the first time I've actually seen a telepresence robot "in the flesh" (as it were) and it was a very interesting experience. What's going to be even more interesting is that Steve says he will let me use a Beam to attend a forthcoming Augmented Reality conference. I cannot wait. I will be reporting more about this later.

Today's keynote presentation -- Large-Scale Deep Learning for Building Intelligent Computer Systems -- was given by Jeff Dean, a Google Senior Fellow. I will be writing more about this later (I have to run down to attend the next session in a few minutes). Suffice it to say that Jeff really opened my eyes with regard to all sorts of concepts I never even knew existed. When he opened the floor for questions at the end of his talk, I simply couldn’t hold myself back.

The view of the microphone seconds before I posed my question
(Source: Max Maxfield /

What did I ask? I'll have to tell you later. The best line I've heard so far came from Bruce Daley from Tractica in his talk How Deep Learning is Enabling Computer Vision Markets when he said: "You can't be too thin, too rich, or have too much data." He was, of course, talking about the data sets required for training neural networks. This is another topic we will be exploring in more detail when I have more time.

Next up was Chris Rowen, PhD, FIEEE, and CTO of the IP Group at Cadence Design Systems. At the end of his session -- The Road Ahead for Neural Networks: Five Likely Surprises -- Chris announced the new Cadence Tensilica Vision P6 digital signal processor (DSP). This is now Cadence’s highest-performing vision/imaging processor, and it extends the Tensilica product portfolio further into the fast-growing vision/deep learning applications areas. I'm looking forward to seeing some demonstrations of this technology later today.

For the moment, however, I must bid you adieu as I race downstairs to attend the next session. Watch this space for ongoing developments...

Loading comments...