Multilayer perceptron with on-chip learning using stochastic ratio pulse arithmetic

The necessity of the hardware implementation of neural networks is discussed. However, it is difficult to realize compact operators into a small silicon area. Therefore we used the ratio pulse method in stochastic arithmetic, and the representation of the ratio of ones and zeros in its random pulse stream. The ratio method has the advantage that it reduces the size of operators in neural networks. This paper proposes learning algorithms, using ratio pulse arithmetic, for an on-chip neural network with circuits.

[1]  Yasuji Sawada,et al.  Functional abilities of a stochastic logic neural network , 1992, IEEE Trans. Neural Networks.

[2]  Naleih M. Botros,et al.  Hardware implementation of an artificial neural network , 1993, IEEE International Conference on Neural Networks.

[3]  Toshiyuki Furuta,et al.  Neural network LSI chip with on-chip learning , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[4]  Soo-Ik Chae,et al.  Neural network implementation using new pulse arithmetic , 1994 .

[5]  Soo-Ik Chae,et al.  Ratio pulse arithmetic for radial basis function network , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[6]  Y. Suzuki,et al.  Digital systems for artificial neural networks , 1989, IEEE Circuits and Devices Magazine.

[7]  G. S. Eisman.  Frequency based computation in neural networks , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[8]  Max Stanford Tomlinson,et al.  A digital neural network architecture for VLSI , 1990, 1990 IJCNN International Joint Conference on Neural Networks.

[9]  Saul B. Gelfand,et al.  Analysis of gradient descent learning algorithms for multilayer feedforward neural networks , 1991 .

[10]  Soo-Ik Chae,et al.  A study on the stochastic computation using the ratio of one pulses and zero pulses , 1994, Proceedings of IEEE International Symposium on Circuits and Systems - ISCAS '94.