Quantum Bayesian Networks

March 15, 2012

Judea Pearl Wins Turing Award For Bayesian Networks

Filed under: Uncategorized — rrtucci @ 6:46 pm

Check out

A Turing Award for Helping Make Computers Smarter
By Steve Lohr (New York Times, March 15, 2012)


Google search, I.B.M.’s Watson Jeopardy-winning computer, credit-card fraud detection and automated speech recognition.

There seems not much in common on that list. But it is a representative sampling of the kinds of modern computing chores that use the ideas and technology developed by Judea Pearl, the winner of this year’s Turing Award.

Dr. Pearl, 75, a professor at the University of California, Los Angeles, is being honored for his contributions to the development of artificial intelligence.

In the 1970s and 1980s, the dominant approach to artificial intelligence was to try to capture the process of human judgment in rules a computer could use. They were called rules-based expert systems.

Dr. Pearl championed a different approach of letting computers calculate probable outcomes and answers. It helped shift the pursuit of artificial intelligence onto more favorable terrain for computing.

Dr. Pearl’s work on Bayesian networks — named for the 18th-century English mathematician Thomas Bayes — provided “a basic calculus for reasoning with uncertain information, which is everywhere in the real world,” said Stuart Russell, a professor of computer science at the University of California, Berkeley. “That was a very big step for artificial intelligence.”

Of course, I believe Bayesian networks are highly relevant to quantum computing and quantum Shannon Information theory. Quantum Mechanics is, after all, just another probability theory.

I first heard of Bayesian Networks from Bill Gates, founder of Microsoft.

Okay, Okay, I don’t really know the guy. I don’t even know his chauffeur. Nevertheless, there is some kernel of truth, this time at least, to my statements.

Since that day, forever etched in my memory, when I overheard Bill Gates’s boffo remarks about Bayesian Networks, I have been a devoted fan of them, especially of their quantum version. My very first paper in ArXix is entitled, appropriately enough, “Quantum Bayesian Nets”. My understanding of q-b-nets has increased quite a lot since that paper. Many of my subsequent papers have dealt with q-b-nets (better than cabinets). Sometime ago, I wrote a Mac application called “Quantum Fog” that does quantum Bayesian networks. (It uses algorithms of exponential complexity so it is only intended for pedagogical purposes. ) And of course, this blog is named after qbnets, and many of its post are about the very subject.

Addendum: Above, I only mentioned Judea Pearl’s work on Bayesian networks. More recently, Pearl has also written some papers and a book on his own theory of causality (which is an extra structure built on top of the foundation of Bayesian networks). However, to date, his theory of causality (“causality calculus”) has been used by others MUCH less frequently than his previous work on Bayesian networks, especially in industrial computer applications.



  1. Scott Aaronson mentioned on his blog that he is Daniel Pearl’s dad. Had no idea. Terribly sad.

    Comment by Henning Dekant — March 16, 2012 @ 3:37 am

  2. Yep. A well known fact among bayesian network freaks like me.

    Comment by rrtucci — March 16, 2012 @ 9:58 am

  3. Do you know of any fruitful cross-pollination between bayesian networks and formalized fuzzy logic?

    Comment by Henning Dekant — March 16, 2012 @ 3:51 pm

  4. Bayesian network advocates do the sign of the cross when someone mentions fuzzy logic, as if someone had mentioned the devil, or the evil eye. They think fuzzy logic is ad-hoc crap that leads to contradictions, just like rule-based systems lead to contradictions.

    (b. nets are not perfect though. For one thing, using them seems to require calculations of exponential complexity).

    By the way, Google is a great advocate of bayesian networks.

    Comment by rrtucci — March 16, 2012 @ 6:07 pm

  5. Interesting. The way fuzzy logic is used in controller applications is certainly very ad-hoc and quick fix like. On the other hand contradictions are handled just fine – simply assign a 1/2 truth value to the statement and its negation 🙂

    It’s brilliant really, as it allows for the machine equivalence of human cognitive dissonance. Obviously without such a common trait no AI will ever pass the Turing test.

    Comment by Henning Dekant — March 17, 2012 @ 3:01 am

  6. Being a physicist, what most fascinates me about b. nets is their connection to quantum mechanics. Their connection to AI fascinates me too, but it’s less physicky, so I’m programmed to love it less 🙂

    Comment by rrtucci — March 17, 2012 @ 11:42 am

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Create a free website or blog at WordPress.com.

%d bloggers like this: