论文信息 - Rapid, supervised training of a two-layer, opto-electronic neural network using simulated annealing

Rapid, supervised training of a two-layer, opto-electronic neural network using simulated annealing

Abstract We have demonstrated for the first time the supervised training of a two-layer opto-electronic neural network using simulated annealing. Our training strategy only accepts solutions which are tolerant to noise in the network. Any 3-bit logic function can be reproduced with training times of a few tens of seconds.

N. M. Barnes | A. W. O'Neill | David Wood

[1] R. P. Webb,et al. High speed opto-electronic neural network , 1990 .

[2] Bernard Widrow,et al. 30 years of adaptive neural networks: perceptron, Madaline, and backpropagation , 1990, Proc. IEEE.

[3] M P Dames,et al. Efficient optical elements to generate intensity weighted spot arrays: design and fabrication. , 1991, Applied optics.

[4] M.A.Z. Rejman-Greene,et al. Planar 3*3 array of GaInAs/InP MQW surface optical modulators grown by gas-source MBE , 1988 .

[5] David Wood. Training high-speed opto-electronic neural network hardware , 1991 .

[6] R. P. Webb,et al. 16-channel parallel optical interconnect demonstration with an InGaAs/InP MQW modulator array , 1990 .