sfleming@cs.strath.ac.uk (Stewart T Fleming IE87) (01/10/90)
One of my lecturers in Signal Processing recently made the rather interesting statement that : "if you can calculate a correlation matrix for a stochastic process, then you know everything about that process". An n by n correlation matrix contains the autocorrelations (dependencies) of n stochastic processes. An autocorrelator can be very easily constructed from simple electronic components (integrator/threshold). If anyone wants an alternative to the current "Can Machines Think" argument, then they might like to develop the idea of the (human) brain as a large autocorrelator of stochastic processes sourced from the senses, noting that the autocorrelations between processes only become valid as the number of observations made tens to infinity. Stewart Fleming (Undergraduate) 4th Year Information Engineering University Of Strathclyde. Disclaimer : I know nothing at all. I just ask questions