Accidental Blogger

Visual computations tend to be generically hard and resource intensive. For example, our state-of-the-art algorithm for yaw detection in flight apparently needs Moore’s Law engorged Intel Cores to do anything useful. And yet mere flies seem to manage the number-crunching just fine…The trick at least in this instance seems to be that flies aren’t using our stupid state-of-the-art algorithm. Theirs is vastly nicer:

By turning the brain cell activity underlying fly eyesight into mathematical equations, researchers have found an ultra-efficient method for pulling motion patterns from raw visual data.[…]

“We can build a system that works perfectly well, inspired by biology, without having a complete understanding of how the components interact. It’s a non-linear system,” said David O’Carroll, a computational neuroscientist who studies insect vision at Australia’s University of Adelaide. “The number of computations involved is quite small. We can get an answer using tens of thousands of times less floating-point computations than in traditional ways.”[…]

The researchers’ algorithm is composed of a series of five equations through which data from cameras can be run. Each equation represents tricks used by fly circuits to handle changing levels of brightness, contrast and motion, and their parameters constantly shift in response to input. Unlike Lucas-Kanade, the algorithm doesn’t return a frame-by-frame comparison of every last pixel, but emphasizes large-scale patterns of change. In this sense, it works a bit like video-compression systems that ignore like-colored, unshifting areas.

There’s probably some generic lesson here – the set of efficient solutions to any problem could well be vastly larger than the subset which makes use of well-separated, clear, extensible cases, modular reasoning, simple and readily debugged steps and the like, which subset in turn is probably larger than that of solutions people tend to come up with. Nature, with no particular need for comprehensibility or ease of maintenance, can happily brute-force out methods that work rather well while making no sense.

I’m sure that’s overstated – eyes work in quite sensible ways – while R&D and further inquiry will doubtless clarify what now seems inexplicable because of novelty. Still, I shouldn’t be surprised if many mind-tricks turned out to be like this – no human smart enough to do the amazing thing would have the cheek to do it this maddening, meaningless, hideous way.