Numenta HTM Challenge 2015 Results
Earlier this month, we ran a contest for our developer community to use our algorithms on real world problems. The contest, called the HTM Challenge, was conceived by Matt Taylor, our community flag-bearer, as a way to enable the community to invest more time in trying to build an application than they can in a weekend Hackathon.
Overall, we were pleased with the result, with 14 entries submitted. Each entrant was required to make a short video demonstrating the application. You can see them all here:
It’s interesting to note that the #1 and #2 winners are both in the field of transportation, one detecting car traffic anomalies and the other relative to aviation anomalies, both using our geospatial encoder.
To me, however, the most notable result of the contest was to realize the breadth of the type of applications that can be addressed with HTM. Although these were just demonstration programs, not ready for commercialization, they offer support to the notion of a generalized machine intelligence approach. Whether the application was predicting traffic, tracking heart monitors, understanding natural language, or generating music, the exact same algorithm is at play, just like in your brain.
We often talk about how our biologically derived technology is extremely flexible and applicable to many different problems. It’s nice to see this reinforced with the wide variety of submissions. Thank you to everyone who participated in this challenge. We look forward to seeing what future applications will be built on HTM.
See photos from the event held in Redwood City here.