Last night, while many of you were coming up with new strategies for approaching this contest, Ned was busy looking at past submissions. Take a look at his thoughts in the Mid-Contest Analysis. There is a Japanese version available as well.
By
Helen Chen
Helen is part of the MATLAB Central team. She spends her days hanging out with really cool folks online at MATLAB Central and at MathWorks HQ in Natick.
Thanks for the analysis, Ned. Very cool stuff.
Also, I don’t think I’ve mentioned this elsewhere yet, but I really like all the new contest machinery that was put in place this time around. I appreciate the effort that the team has put into the contest.
I agree. Very nice work with all the new features you and your team have added since the last contest.
Also, I do believe that the graphs at the end on the Stats page are not getting updated along with the other stats.
Thanks for making this fun and extremely educational contest !
Thanks, Ned, for picking out my Super Slim entry. I couldn’t resist dipping my toe in again. A reference framework was precisely the idea, though I had hoped that the JIT compilation of 2010a would’ve made it a bit faster – I was testing on the copy of 2007b I have at home. Rather than computing exact mean values for queryLimit regions, the regions overlap a lot, so there are many more of them. This increases resolution, but at the cost of a harder inference problem which requires some regularization.