Quantum Bayesian Networks

February 19, 2011

I For One Welcome Our New Probabilistic Computer Overlords

Filed under: Uncategorized — rrtucci @ 4:44 pm

On TV broadcasts aired on Feb. 14, 15 and 16 of 2011, IBM’s Watson computer played the trivia game of Jeopardy with the all-time best human players, Ken Jennings and Brad Rutter, beating them by a wide margin.

In his final answer, Ken Jennings wrote “I for one welcome our new computer overlords”

Jeopardy-relevant trivia about the origins of this phrase:

(My source and more information here): Operation Overlord was the code-name given to the Battle of Normandy, June 1944, the World War II battle on the west coast of France that launched the Allied Force invasion of German occupied Western Europe.

Arthur C. Clark’s 1953 sci fi novel “Childhood’s End” revolves around an alien race known as the Overlords that takes over the human race.

The 1977 film adaptation of the H.G. Wells short story “Empire of the Ants” uses the phrase “I, for one, welcome our new insect overlords”.

In the 1994 episode of The Simpsons titled, “Deep Space Homer”, news announcer Kent Brockman believes that the Earth is being invaded by giant space ants. He says “One thing is for certain: there is no stopping them; the ants will soon be here. And I for one welcome our new insect overlords. I’d like to remind them that as a trusted TV personality, I can be helpful in rounding up others to toil in their underground sugar caves.”

Much has been said and will be said in the future about the Feb. 14-16 Jeopardy event. Everyone interprets the event slightly differently, depending on their own experiences, interests, and world view.

To me, it was a spectacular, dramatic, breath-taking, amazing, awesome event. A humbling moment for carbon based life-forms. A great technological and scientific achievement in fast natural language processing. A grand synthesis of prior computer hardware and software inventions. A giant publicity coup for IBM.

(1)Querying a computer in natural language à la Star Trek
(2)Having that computer hypothesize multiple answers to that query, and then cross-check those answers and rate its confidence on them, basing those answers, not on a laboriously curated data base, but on an unstructured data base that is written in messy natural language.
Watson has jumped well over these two major AI hurdles with great panache (displaying super-human response times and success rates, performing high up in the “winner’s cloud”).

I see Watson as a powerful vindication of the idea of using probabilistic techniques for AI (Artificial Intelligence) and machine learning. Since I’m so interested in quantum computing, I can’t help pondering about the connections and implications of Watson for quantum computing. The main connection I see is that quantum computers are probabilistic computers too. In fact, QCs are ideally suited for doing quantum mechanics, and quantum mechanics can be viewed as a souped up version of classical probability theory. In my work, I use QCs to do Bayesian networks and Gibbs sampling, two of the most important techniques of probability theory and AI.

I hear the “Raiders of the Lost Ark”, Indiana Jones theme music play in my head every time I think of Watson’s intrepid programmers. I see Watson’s amazing software as a reminder that computers are useless without good software. It seems to me that currently, in the quantum computing field, software development is not being stressed enough.

I see Watson as an inspiration for people, corporations, and nations to do more science (including Quantum Computing Science):

  • For people: This event is sure to inspire countless young people to pursue scientific careers, or at least to learn more science.

  • For corporations: This event may encourage more corporate investment in scientific R&D. The event is a tremendous boost to IBM’s reputation. IBM suddenly doesn’t look so old-fashioned, dowdy, and out-of-touch anymore. Google, Microsoft and Autonomy must be green with envy at this technological and publicity coup in AI by IBM. They are sure to redouble their own, already considerable efforts in the AI field, so as not to be left behind by IBM. Shouldn’t the all mighty Google, king of answering questions, have done this first?

  • For nations: This event sends a clear message to everyone in the world that American science and technology, in particular its computer science and technology, is still among the strongest in the world. America’s reputation is greatly enhanced by this. Other nations will try to emulate us (and also compete against us) in this arena. Such competition should be welcomed, since it will result in significant advancements in science, something that ultimately benefits us all.

Some questions to think about:

  • How significant is this event in the history of AI ?

  • How does Watson work? Describe in detail Watson’s software and algorithms (i.e., the magic behind the curtain).

  • What would be a good incremental next step for Watson and AI? An obvious one is integrating Watson with already existing and quite robust speech recognition technology. Another one is porting Watson technology to non-English languages (grunt work, but consider how useful it would be and how many good jobs would be generated in the writing and utilization of such ports).

  • How soon before this technology shows up in commercial products?

  • How soon before amateur computer programmers (hackers) can look at and tinker with the Watson code, or another computer code with similar capabilities? According to Wikipedia, “Watson was written in both Java and C++[12] and uses IBM’s DeepQA software and SUSE Linux Enterprise Server 11.”

  • Will a supercomputer be necessary for running future versions of Watson? According to Eric Brown, one of the research managers of the Watson project, the first version of Watson ran on a Blue Gene/P, IBM’s biggest supercomputer, but this turned out to be way too slow and was replaced by the current POWER7 cluster architecture, which uses fewer cores than a Blue Gene/P, but those cores run 4 times faster. (Blue Gene/P has 4,096 cores that run at 850 MHz, whereas Watson uses a POWER7 cluster with 2,880 cores that run at 3.55 GHz).

    It’s also quite telling that while the computer program Deep Blue that beat Gary Kasparov in 1997 required a specialized supercomputer to run, nowadays one can purchase computer programs like Rybka, Deep Fritz or Deep Junior which can consistently beat Deep Blue and require only a single personal computer to run. These newer programs use much smarter, more efficient algorithms than Deep Blue.

Some References

  1. Smarter than you think. What Is I.B.M.’s Watson? By Clive Thompson (New York Times, June 16, 2010)

  2. How Watson works: a conversation with Eric Brown, IBM Research Manager, by Amara D. Angelica (January 31, 2011)
  3. Wikipedia article Watson_(artificial_intelligence_software)
Advertisements

2 Comments »

  1. Watson? Commercial – Not Super – Computer…

    Now that IBM’s Watson has pounded the ideal human Jeopardy competitors into a fine slurry, let’s take stock. Our human proxies took their ass-kicking in good spirits, with Ken Jennings writing on his ‘Final Jeopardy’ card, “I for one welcome ……

    Trackback by Gad-tech.com — February 19, 2011 @ 8:00 pm

  2. Excellent blog post by Tony Pearson (az990tony)
    IBM Watson – How to build your own “Watson Jr.” in your basement (Feb 18)

    Comment by rrtucci — February 25, 2011 @ 5:33 pm


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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

Create a free website or blog at WordPress.com.

%d bloggers like this: