by Numenta Inc., Hierarchical Temporal Memory (HTM) is a
new type of computing technology modeled on the structure and
operation of the neocortex, a thin sheet of brain tissue that
covers the outer layer of the brain and is said to be the seat of
HTM system is not programmed in the traditional sense; instead it
is trained. Sensory data is applied to an HTM system and
automatically discovers the underlying patterns in the sensory
input. HTM's learn what objects or movements are in the world and
how to recognize them, just as a child learns to identify new
objective for this project was to learn how an HTM system worked
and develop an application that could be used to add machine
intelligence to systems used in building automation and process
project was divided into two parts; first learn how to program an
HTM network, and then set up a system of sensors and cameras that
would generate the required data to train the HTM system to
control the security, lights and HVAC system in a building.
early release of the HTM programming platform (NuPIC) was
difficult to learn and could not be adapted to our application.
However the concept of developing a remote site that could be
monitored and controlled over the internet resulted in gaining
industry support and sponsors.
November of this year (2009) an application released by one of
Numenta's partners has demonstrated the use of HTM technology to
process images from video cameras. Recent testing in our research
lab has shown the ability this HTM based technology to provide a
high degree of intelligence in what a video camera is capturing,
equal to or surpassing some of the most expensive video analytics
software on the market.
project has been re-opened and will be included in our Security
and Home Automation project at the Kortright Centre.