What people are saying - Write a review
We haven't found any reviews in the usual places.
Theoretical Aspects of Back Propagation
The Two Wire Analog Back Propagation System
Typical Squash Function and Derivative
9 other sections not shown
activation function approximately assumed assumption Back Propagation algorithm backwards flow boolean functions capacitor chain rule Chapter connected convergence data spikes Discrete Weight algorithm Discrete Weight neural dopj equation error information Error output error signal error value Figures 4.2A Finite Step floating point flow of error forward flow gradient descent Hence hidden layer hidden processing elements inhibitory spike input layer input lines input processing elements learn rate line 82 Linear Threshold Units minima monotonically decreasing multiplication neuron number of bits number of spikes output layer output processing elements output signal output units partial derivative pattern presentation Perceptron performed Poisson distribution presentation of pattern probability problem processing element circuits Pulse Width Modulator Pulsed Error line signal trace single bit weights single weight Spike Model Spike Trains squash function summation synaptic elements systolic array total error traditional Back Propagation utilizing VLSI weight change Weight neural network weight value zero