“AI at the Edge” – coming soon to a theater near you!

OK. Maybe not to a theater.

But most certainly, a live webinar (where you can offer questions during the live Q&A at the end), to be hosted by Avnet on Thursday, Dec. 5th, 2PM EST. (Don’t worry; I’ll send out reminders.)

Why “AI at the Edge”? And Why Now?

You know that I’m mostly a theoretician. (Love ’em equations.) So for me to go over to the dark side – I mean, something practical – you know there’s been a shift.

What happened, simply, is that I got asked. You know how this works – friend is an editor, that sort of thing …

The company sponsoring this edumercial … I mean webinar … is Avnet, which provides a range of devices and solutions, as well as an entire supportive ecosystem. They’ve recently focused on providing AI-based solutions, particularly addressing the IoT world. Here’s one of the quotes they’ve taken from Global market intelligence firm IDC, 2019.

The digital universe is doubling in size every two years with the number of connected devices set to top $40 billion in 2020. Worldwide IoT spending is growing at a double digit rate as well with spending to surpass $1 trillion by 2022.

Avnet is reaching out to their customer base with a suite of information on both AI and IoT.

You know how the best marketing isn’t selling, it’s teaching? It’s providing the best information so that clients can formulate and assess solutions to their problems. That way, when you provide the solution – the relationship is already there. 

Well, that’s what Avnet is doing, and it seems that they’re doing a mighty fine job. (I’m not the only person that they’ve asked to do a webinar, and they’re following up with white papers – the full suite of education offerings.)

So … AI is moving – has moved – from the dev lab to the cloud and now is moving out to the edge. LOTS of AI-devices that are specialized for this, and more on their way. 

 

Your Live Q&A Opportunity

So earlier this week I was cutting the audio track – working with an East-coast-based team (after years of rolling solo, that felt weird! – and strangely good).

We do a sort of tech-rehearsal this coming Monday, and then roll the whole thing live on Thursday, Dec. 5th. That’s rolling live with the pre-recorded audio-over-slidedeck – but then yes, live, you can ask questions and get them answered real-time.

We’ve only got a ten-minute window at the end for the live Q&A, so if you want to submit something, the smartest thing is to email me:

alianna@aliannajmaren.com

That way, I can get your question(s) into the queue, and you’ll have a lot better chance of having your question addressed.

Preview of Coming Attractions (and Recent Releases)

A LOT has happened lately. That’s partly my excuse for not staying in touch with you the way I’ve planned (and you’d come to expect). 

The truth is: I got scared. Nervous Nelly. All just sort of shy and weird. 

And the reason for that is: all of a sudden I actually put more stuff out there … and then just got very uncomfortable with putting out the word that there was more stuff. 

I really can’t explain it better. Just … stage fright, ok? 

So I’m going to make the next few blog posts focused on things that were done one-to-three months ago … as well as moving forward with new stuff. 

Things already published (but not announced):

• A major monograph on Karl Friston’s treatment of variational Bayes (where he applies it to a system enclosed in a Markov blanket); I call this the Rosetta stone monograph because it enables translation between Karl’s work and other, more traditional treatments of variational Bayes, 
• Another big fat monograph, summarizing all the computational results so far on the 2-D cluster variation method (CVM), with Appendix A giving the derivation (in 100 easy-to-follow equation-steps) for the analytic solution. (This analytic solution was published by Kikuchi and Brush in 1967, but they gave the results only … don’t you just love physicists?)

I’ll do a catch-up on these with you shortly. 

Subscribe! Like! Hit the Notify Button!

Yup, You guessed it. I’ve started making YouTubes, and have a YouTube channel. (Can you believe it?)

What really happened is that I got a sort of life-punch-in-the-gut, and the only way that I could resolve it was to get focused on something that I knew was important – but had just been dilly-dallying about for a couple of years – making YouTube vids. 

And why YouTube vids, and why now? 

Because we just learn better when we can combine audio with visual. Our brains just assimilate better that way. And by visual, I mean a really tight slidedeck. (Animated or actual vid is better, but a tight slidedeck is all I can manage right now.) 

So, for example, a 10-minute vid might have about 20 slides (images) in it. 

The upside is that you get a much higher density of information flow. In fact, if I can produce a really good, tightly-edited vid – with LOTS of images illustrating each specific little point – the information density might be equivalent to a 30-minute lecture, where I’m just drawing things on the whiteboard or doing a live class lecture. This is because I can be very focused, and edit out all the little distractions. 

The downside is that it takes a LOT of time to create a quality vid. 

If I were to just record an audio track, that wouldn’t take much time at all. But in my two “experiments” so far, it took about 36 hours (including time for two short naps, and eating only what was in the fridge – not stopping to really prep meals) to create the first 12-minute vid. 

It took me about 18 hours to create the second, 15-minute vid – which started off at 19 minutes and then had to be edited back. (I’m using Camtasia, and YouTube puts a 15-minute limit on video uploads from Camtasia. Weird, but that’s what it is.) That was 18 hours start-to-finish, no naps; working straight through. 

So … I can’t pull this off every week. Even when I calm down, and take more time for production (and am not pressing hard to just get the damn thing done) … well, a weekly YouTube vid will likely be beyond my scope. 

At the same time, now that I’ve got the hang of it, I have a whole lotta vids that I want to produce. So that’s where this coming Christmas vacation is going to be spent.

Just for fun, here are the first two vids: 

YouTube #1: The Transfer Function: 

YouTube #2: Neural Network Mesostructures (simple architectures): 

These first two YouTube vids are deliberately just plain boring. They’re really tech-checks; my getting used to the process. The first stage of the learning curve.

The upcoming series – the one that will be beyond just “director’s cuts” from the Avnet webinar – will focus on the most important thing in deep learning: getting beyond backpropagation. 

More to come. (Back to blogging on a regular basis, so look for this story to continue next week.)

Live free or die, my friend –

AJ Maren

Live free or die: Death is not the worst of evils.
Attr. to Gen. John Stark, American Revolutionary War