robert@aerospace.aero.org (Bob Statsinger) (03/23/90)
In article <4098@zehntel.UUCP> donw@zehntel.UUCP (Don White) writes: > > I wonder how one would go about quantifying the fractional dimension of > an neural net. Hmmmm. > > Don White > Box 271177 Concord, CA. 94527-1177 > zehntel!donw > For unsupervised nets with feedback (eg, BAM, ART), one can use a Lyapunov function as a stability criteria (ie, show that the L function's time derivative is strictly decreasing so that solutions are shown to be strongly stable). This function is presumably an approximation to the complete Lyapunov expansion of the state space of the net. The Lyapunov expansion can be used as a measure of the net's fractal dimension. -- Bob Statsinger Robert@aerospace.aero.org The employers expressed herein are strictly mine and are not necessarily those of my opinion's....uh..er...whatever...