Machine Intelligence · Artificial intelligence

How would your start-up cope with the current AI (mostly Machine Learning) revolution?

Mikhail Gorelkin Principal AI Architect & Developer

June 10th, 2015

Here is what is going on: http://www.forbes.com/sites/anthonykosner/2014/12/29/tech-2015-deep-learning-and-machine-intelligence-will-eat-the-world/

Mikhail Gorelkin Principal AI Architect & Developer

June 10th, 2015

It’s really interesting what you do. It seems it’s a kind of cognitive computing.

Dan Oblinger Founder at AnalyticsFire

June 11th, 2015

Mikhail,

As a researcher and practitioner in the area I am supportive of the enthusiasm inherent in the way you frame your question.

But the reality in 2015 is that this tech is quite finicky stuff.  its hard to make it work, and when it does work it provides modest (but real) advantage over conventional supervised induction.  So I think the idea that if a company is not on the deep learning bandwagon, they will soon be buried is not really accurate.

We are still having difficulty making the stuff be of practical consequence.  Many companies are investing enormous sums based on the promise of the tech, but few have reaped the rewards that prove this bet good.

(Still I am betting with them!  In a winner take all world, small advantages can prove the difference between #1 and #2, which can make all the difference.)


A second related point:

We don't really know how to use deep learning in broader contexts, today in practical application I believe it is almost always used as an upgrade for conventional supervised induction, which is typically limited to narrow prediction or modeling tasks.  Again I am bullish on the value of this, but I have to admit that its effects on an organization can sometimes be limited by this.

--dan

Mikhail Gorelkin Principal AI Architect & Developer

June 11th, 2015

Dan,

I agree with your point: deep learning is perhaps the most difficult technique from ml, and there are still tons of practical problems there including scalability, but there are start-ups that already master it: AlchemyAPI (now part of IBM) for image and text, MetaMind for image and Twitter sentiment analysis, etc.

Deep learning is only part of this revolution; it includes also other ml techniques, text mining / nlp, cognitive computing, etc. My point is: the most sophisticated engineering gives us a really good chance to compete on the current market.


Dan Oblinger Founder at AnalyticsFire

June 11th, 2015

I love those startups, I am rooting for them.  I have even used the AlchemyAPI... still the skeptic in me would classify them as promising but unproven.  They have succeeded in getting money, building a product, and even getting acquired.  

Still delivering large value to some end customer is a item that is TBD.

but of course the supervised induction tech developed in the 1990s  HAS now past that rubicon,
it is used as back end optimization technology in thousands of commercial contexts.

good discussion,
--dan

Mikhail Gorelkin Principal AI Architect & Developer

June 11th, 2015

Google's voice recognition is already in production and is based on deep learning. They even extended their capacity in DL by acquiring DeepMind the last year. Beyond Socher's start-up are resources of Stanford. So I think the results would be very impressive as well.


About AlchemyAPI: it seems IBM is going to extent its cognitive framework based on IBM Watson (mostly nlp primitives + automated reasoning) with AlchemyAPI's feature engineering based on deep learning. The game is going from computational primitives to higher levels of machine intelligence.

Karl Schulmeisters CTO ClearRoadmap

June 10th, 2015

By using it