I gave a presentation at work today to an audience of probably 100 to 150 people, describing my project of the last week.
The project involved genetic algorithms for auto-tuning configurations, and I almost put the “Kanye – Gay Fish” song from the “South Park” episode “Fishsticks” into the presentation:
But the Auto-Tune comes in a bit too late. It was a 5-minute presentation, and I could only spare a few seconds for the Auto-Tune joke. Instead I went with Auto-Tune the News:
I think these choices further prove how cool my job is.
I’m pretty satisfied, both with my presentation, which I ended up calling “Auto-Tune the Config,” and my project in general. It’s not quite ready for production yet, but it’s going to cut down on hours of work every month.
Interestingly, when Turing Award laureate Leonard Adleman was asked on stage at the ACM Turing Centenary Celebration what his least favorite kind of algorithm was, he said genetic algorithms, exactly what I was using in my project.
Adleman said using genetic algorithms usually amounts to saying “I don’t understand the problem, but I hope the computer will.” Sure, I admit to not understanding my problem here, how one value influences all the other ones in a rapidly changing multidimensional parameter space. What I care about is finding a solution, one that involves little work on my part.