Sunday, November 27, 2011

The hidden mechanics of humor?

Humor is the embodiment of a latent method that discovers false inference in the brain. What an intriguing thought! Does this imply that a logician and a comedian never mix up? I simply couldn't resist to re-quote the Slashdot post [1]:

"The sense of humor is a ubiquitous human trait, yet rare or non-existent in the rest of the animal kingdom. But why do humans have a sense of humor in the first place? Cognitive scientist (and former programmer) Matthew Hurley says humor (or mirth, in research-speak) is intimately linked to thinking and is a critical task in human cognition because a sense of humor keeps our brains alert for the gaps between our quick-fire assumptions and reality. 'We think the pleasure of humor, the emotion of mirth, is the brain's reward for discovering its mistaken inferences,' says Hurley, co-author of Inside Jokes: Using Humor to Reverse-Engineer the Mind. With humor, the brain doesn't just discover a false inference - it almost simultaneously recovers and corrects itself. For example, read the gag that's been voted the funniest joke in the world by American men. So why is this joke funny? Because it is misleading, containing a small, faulty assumption that opens the door to a costly mistake. Humor is 'when you catch yourself in an error, like looking for the glasses that happen to be on the top of your head. You've made an assumption about the state of the world, and you're behaving based on that assumption, but that assumption doesn't hold at all, and you get a little chuckle.'"

[1] http://science.slashdot.org/story/11/11/27/037237/the-science-of-humor

Monday, November 21, 2011

Get a taste of Stanford: Online lectures 2012

These days I'm having too many irons in the fire. To give you some hints, there's more coming which sheds lights on turning non-negative matrix factorization into an embarrassingly parallel problem (plus avoiding the painful hadoop overhead [1]) and something new which hopefully combines machine learning methods with agile development methodologies to solve a well-known ancient problem in software engineering.

Anyhow, to keep this blog alive, here's the latest from Stanford. Anyone who has listened to the initial three courses (Database System, Machine Learning and Artificial Intelligence) surely has enjoyed their quality and will enlist again. They are back with an incredible list of free interactive video lectures on machine-learning:


There is also a set of basic introductions to CS, SaaS and Entrepreneurship. The lectures are splitted into weekly video sessions and optional quizzes or programming exercises.

The offerings will start this January, so don't miss your spot and sign up now!

[1] https://mpi-inf.mpg.de/~rgemulla/publications/rj10481.pdf