I've been developing an A.I. engine that is loosely based on Jeff Hawkin's theories of how the human neo-cortex works. The theories are based on Hierarchical Temporal Memory (HTM). The machine absorbs airline passenger data and learns in a manner analogous to human development. After training it then predicts whether
future passengers will show up for their flight. In the general population roughly 3.46% of passengers don't show up for their flights (no-shows). My engine is producing lists of passengers that will no-show that are consistently 30-35% correct. The impetus for me to finally reach out to another group is the challenge of assigning significance to patterns. In order to break the 30-35% level I'm testing a new significance module that attemps to apply some of the theories of Ockham, Epicurus and Solomonoff to the HTM model.
I want to convert my engine to a stock market based "world". I don't have any contacts in this world and would like to partner with someone that could provide data feeds.