Quantum Bayesian Networks

February 3, 2015

Scott Aaronson Bites Seth Lloyd’s Ear

Filed under: Uncategorized — rrtucci @ 4:04 pm

Yesterday was an eventful Groundhog Day.

Check out

Jimmy the Groundhog turns on the mayor
(Feb 2, 2015, Chicago Tribune)

excerpt:

SUN PRAIRIE, Wis. — A mayor in Wisconsin has learned a valuable lesson about his city’s weather-predicting critter: Don’t get too close.

Jimmy, the official groundhog in Sun Prairie, bit the mayor’s ear during a Groundhog Day celebration Monday. The groundhog’s handler was holding the animal next to Mayor Jonathan Freund’s face when it promptly bit down on his ear. Freund flinched but went on with his declaration that Jimmy had predicted an early spring.

Will the groundhog see its shadow? It’s a question asked of the country’s most famous critter, Punxsutawney Phil, every Feb. 2 in front of anxious crowds desperate to know whether spring is around the corner or if six more weeks of winter must be endured.

The Groundhog Day celebration became even more precarious when Jimmy’s handlers, Jerry and Maria Hahn, said the mayor had gotten it wrong and that there would be six more weeks of winter. Then the city later issued a statement saying only the mayor can translate Jimmy’s prediction.

Next, check out Scott Aaronson’s newest essay, which occurred on the same day, but 1000 miles away, in Boston, MA:

Quantum Machine Learning Algorithms: Read the Fine Print, by Scott Aaronson

Do read the fine print—he really means it: the whole paper is written in 9 point font.

Is Scott burying Lloyd or praising him? His essay does both.

This is Scott’s argument, his impeccable logic, in my own words: Seth Lloyd and his minions have proposed an algorithm for doing AI with a quantum computer. Their dodo bird algorithm could, under certain totally impractical circumstances, fly. To put it differently, their dodo bird is a very promising “template” for the evolution of flying birds, a veritable “mini-revolution” in aviation history. In fact, their dodo bird flies almost as well as … a chicken.

In this video, Matthias Troyer also points out that the AI algorithm proposed by Lloyd et al is virtually useless in its present form.

Scott doesn’t deign to mention the existence of other algorithms for doing AI with a quantum computer. For example, the annealer algorithms used by D-Wave/Google/NASA, or gate model algorithms (such as one proposed by Microsoft people and one proposed by me) based on Grover’s algorithm.

Related blog posts:

ground-hog-day-2015
ground-hog-lloyd
(click icons to enlarge them)

5 Comments »

  1. My understanding is that the Lloyd algo doesn’t really learn all that well. That’s why I see a bright future for this approach. After all amidst all the focus on Artificial Intelligence it is often overlooked that in order for machines to appear more human Artificial Stupidity is at least as important, if not more so. AS and the more rigorous field of ASS (Artificial Super Stupidity) may greatly benefit from this quantum speed-up.

    Finding the wrong answers – exponentially faster! I think Douglas Adams would have approved.

    Comment by Henning Dekant — February 10, 2015 @ 5:41 am

  2. Good point Henning. The Seth Lloyd et al AI algorithm is more likely to solve problems the human way, by starting wars. We can safely predict that the folks at DARPA will love this feature (not a bug) of the algorithm, and will want to invest billions of dollars in it.

    Comment by rrtucci — February 10, 2015 @ 3:46 pm

  3. Hello Bob: Here is a recent interesting talk at Google, by Dr. Daniel Lidar of USC, about QC in general and about D-Wave in particular. Thanks and have a good day.

    Comment by Sol Warda — February 21, 2015 @ 12:40 am

  4. Hi Sol, Thanks very much for the link. Watched the whole thing. Found it very interesting.

    Artiste-qb.net will be neutral about the debate between gate-model versus annealer. We will write software for both models, and for kitchen sinks too if we have to. As long as it’s not a kitchen sink designed by Seth Lloyd.

    Comment by rrtucci — February 21, 2015 @ 5:15 am

  5. This is funny. The people that wrote the Quipper software think they can do Lloyd et al’s algorithm for solving systems of linear equations, and doing machine learning, but they will need a mere circuit depth of 10^25. Hehehe

    http://arxiv.org/abs/1505.06552

    Comment by rrtucci — May 26, 2015 @ 7:36 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: