Machine Learning Engine
Behavioral Recognition Systems, Inc., USA
Behavioral Recognition Systems is a technology start-up based in Houston, Texas specializing in Behavioral Analytics™ -- on-the-fly behavior analysis of video-feed content using AI techniques and technologies.
We began developing our Lisp-based Machine Learning Engine in 2006 using the best of the open source Lisp compilers then available. After several months we decided that we really needed a more stable, better supported compiler and began looking for other solutions.
After speaking with several commercial Lisp vendors we decided that LispWorks provided the best price-performance combination available. Nothing in the ensuing two years has given us any reason to regret that decision. Updates have been timely and benign. On the rare occasions when we have found bugs, they have been fixed rapidly. When we thought we needed additional features or needed the compiler to work differently that it did, LispWorks technical support suggested solutions within the existing capabilities of the product to solve our problems. If we persisted in asking for extensions to the product, then the extensions were added. While our more "hardcore" developers still do use Emacs as a development environment, the LispWorks IDE has proven to be quick, convenient, and so powerful that even they resort to it when debugging problem code.
The source code for the Machine Learning Engine has now grown to more than 185 thousand lines of Lisp code, yet LispWorks compiles it all from a clean start into a tight native binary in less than 90 seconds. The engine is massively multi-threaded (typically 30-35 threads), and large (typically 2-3GB RAM), yet the generated code is so stable that the product can run for literally months at a time.
The Machine Learning Engine, entirely implemented in LispWorks, is the core technology underlying our AISight™ product.
We could never have come so far, so quickly, without LispWorks.
Wesley Cobb, Ph.D.
Lead Scientist, Research and Development
Behavioral Recognition Systems, Inc.