IJCNN International Joint Conference on Neural Networks: July 8-12, 1991, Washington State Convention & Trade Center, Seattle, WA, Volume 1IEEE, 1991 - Neural circuitry |
From inside the book
Results 1-3 of 81
Page 50
... nodes are connected to the output nodes via weights wlm . The output layer nodes use the sigmoidal transfer function , y2m ( y1 ) = L 1 1 + e where , z = ( y11 * Wlm ) 1 = 1 ( 2 ) For two category ( defect / anomaly ) differentiation ...
... nodes are connected to the output nodes via weights wlm . The output layer nodes use the sigmoidal transfer function , y2m ( y1 ) = L 1 1 + e where , z = ( y11 * Wlm ) 1 = 1 ( 2 ) For two category ( defect / anomaly ) differentiation ...
Page 154
... nodes infinity . Then , one by one , compare each node label except that at node ' 1 ' with the sum of the label of node ' 1 ' ( i.e. , 0 ) and the direct distance from node ' 1 ' to the node in question . The smaller of the two numbers ...
... nodes infinity . Then , one by one , compare each node label except that at node ' 1 ' with the sum of the label of node ' 1 ' ( i.e. , 0 ) and the direct distance from node ' 1 ' to the node in question . The smaller of the two numbers ...
Page 325
... nodes ) , ii ) to compute or adapt their routing strategies on line by measuring local variables and exchanging a small amount of messages with neighbouring nodes . The first requirement leads to regard routing nodes as the cooperating ...
... nodes ) , ii ) to compute or adapt their routing strategies on line by measuring local variables and exchanging a small amount of messages with neighbouring nodes . The first requirement leads to regard routing nodes as the cooperating ...
Contents
A Neurocomputational Approach | 11 |
Particle Tracking by Deformable Templates | 7 |
Comparison of Perceptron Training by Linear Programming and by the Perceptron Convergence | 8 |
199 other sections not shown
Common terms and phrases
accuracy algorithm analog annealing applications approach approximation architecture artificial neural network associated keywords backpropagation binary Boltzmann machine chip circuit classifier coefficients components compression computation convergence corresponding data set defined deformable templates detection detector diagnosis digits document dynamic thesaurus error estimate feature map feedforward Figure filter gradient gradient descent hidden layer hidden units Hough transform IEEE implementation input layer input pattern input vector Kohonen learning algorithm learning rule linear matrix measure method minimize multilayer perceptron neural net neurons noise nonlinear obtained operation optimal output layer output unit Parallel Distributed Processing parameters pattern recognition perceptron performance pi-sigma network pixels problem Projection Pursuit proposed recurrent neural networks sample segment self-organizing sensor shown sigmoid function signal simulated annealing simulation solution summing units techniques test set threshold tracks training set update values valve plate variables vector quantization weight vector