|
PureBytes Links
Trading Reference Links
|
Cameron:
Nueral nets are efficient algorithms that don't do every iteration of linear
programming's full factorial analysis. But you don't know what the
software's doing and have to trust it blindly. No matter what data you feed
it for inputs (like you indicators from a chart), you can't give them
contraints of 25<x<35 because the software uses 'proprietary' algorithms
like 'nearest neighbor' and other efficient Operation Research algorithms.
Linear is slow and has limits in solutions to nonlinear problems. But linear
is all we truly understand and therefore translate all nonlinear solutions
thru linear logic (central limit theorem or bell shaped distribution curve).
Proprietary in reality means they don't understand what works sometimes and
don't want others to scrutinize what they don't understand themselves.
Phil
|