论文信息 - Artificial neural networks as building blocks of mixed signal FPGA

Artificial neural networks as building blocks of mixed signal FPGA

Ever since the deployment of FPAAs, efforts are on the way to minimize the silicon area to realize an arbitrary system. A relatively new concept which has been tested and tried in this direction is the use of Artificial neural networks (ANNs) as Configurable Analog Blocks (CABs). Conventional ANNs however suffer with lengthy training period. In this paper ANNs with differential feedback technique are explored. It has been found out that they perform better than the conventional ANNs.

K. S. Gurumurthy | R. Manjunath | R. Manjunath | K. Gurumurthy

[1] Shun-ichi Amari,et al. Information geometry of the EM and em algorithms for neural networks , 1995, Neural Networks.

[2] Laurene V. Fausett,et al. Fundamentals Of Neural Networks , 1993 .

[3] Ehl Emile Aarts,et al. Simulated annealing and Boltzmann machines , 2003 .

[4] Shun-ichi Amari,et al. Natural Gradient Works Efficiently in Learning , 1998, Neural Computation.

[5] Shun-ichi Amari,et al. Methods of information geometry , 2000 .

[6] S. Amari. Differential Geometry of Curved Exponential Families-Curvatures and Information Loss , 1982 .