Congratulations!

Daphne Koller just won a MacArthur “genius” grant. Daphne's research focuses on decision making in uncertain environments, combining techniques from game theory, probability theory, learning theory, classical artificial intelligence, and theoretical computer science. [Thanks, Lance!]

Now what was this again about women and theoretical machine learning?

Comments

I understand the argument you are trying to make but I don't think it can be proven simply by pointing out theoretical researchers who happen to be women.

The original articles contention was that women tend to shy away from theory towards applications for whatever reasons. So we would need to look at relative ratios of theory vs. applied and see if there is some pattern that needs to be explained.

In my (American top 10) CS department among the women faculty 2 of them have a mainly theoretical focus and 4 are mostly applied. Interestingly enough among the younger female graduate students here there seems to be a stronger interest in theory than applications.

But its really hard to draw any conclusions since the numbers are too small.

Actually now that I think about it, I am more convinced that this effect is probably small or non-existent. People seem to research stuff that interests them and I certainly have not heard anyone "hiding behind massive datasets." If anything they want to avoid tedious experiments so they can graduate faster!

[Jeff Erickson]

I agree—enumerating a dozen examples doesn't prove anything. As you say, given the small sample size, it's hard to draw any conclusions.

I see a horrible gender imbalance in my department among both students (roughly 5%, perhaps more among the grad students) and faculty (5/50 tenure track). I see a similar imbalance at conferences in my own research areas, algorithms and computational geometry. But I don't see a WORSE imbalance in my area than in computer science in general, nor a preference in the women in my field to lean toward applications. If anything, it's the opposite. Coming up with that list of female theoreticians was easy.

Hence my shock at Ms. Wallach's suggestion.

I also don't see a significant difference in the fraction of women between computer science and electrical engineering (more applied) or mathematics (more theoretical). If anything, there seem to be slightly MORE female mathematicians than engineers. But I don't have any hard data, and being married to a mathematician may have inflated my exposure to math babes.

A third of the female faculty in your department prefer theory?! That's quite impressive, actually. What's the ratio among MALE faculty? (I'd say about 25% of the tenure-track faculty in my department are mainly theoretical, but none of the female faculty.)

"If anything they want to avoid tedious experiments so they can graduate faster!" Heh heh. Riiiight. If only it were that simple!

I just looked at the webpages and a very rough and probably wrong count has about 17 theory faculty and 66 faculty overall. So it seems the size is roughly 25% as with your department.

As far as the experiments, this is just the conventional wisdom among the students here. They think that a theory PhD is faster because once you get some kind of preliminary results you are supposedly guaranteed to graduate soon afterwards. After all, they can't really say "No graduation for you unless you prove much stronger results!"

Whereas with building systems and doing experiments they can and do impose more inflexible criteria because they know you can meet them.

That said, I am not sure if this thinking is correct. The main factor seems to be how quickly the student finds a good topic and can make significant progress on it. Plenty of applied people seem to get out pretty fast and some theoretical people wander for years without focus.

I guess like the women thing, this one is also a bit hard to verify without more data.