DocumentCode :
2958899
Title :
VHDL-AMS behavioral model of an analog neural networks based on a fully parallel weight perturbation algorithm using incremental on-chip learning
Author :
Michel, J. ; Herve, Y.
Author_Institution :
Univ. Louis Pasteur, Strasbourg, France
Volume :
1
fYear :
2004
fDate :
4-7 May 2004
Firstpage :
211
Abstract :
An analog neural network VHDL-AMS model is developed to analyze the performances of an IC architecture associated with a learning algorithm. We compare here an electrical simulation of a current mode architecture dedicated to deep submicronics technologies with a formal MATLAB model. This comparison allows researching suitability between architecture and algorithm to optimize the learning speed versus the classification precision. It allows also to study the robustness of architecture versus electrical noise, component dispersions or memory loss and the robustness of an algorithm versus noise in the data´s.
Keywords :
analogue integrated circuits; hardware description languages; integrated circuit noise; neural nets; perturbation techniques; IC architecture; MATLAB model; VHDL-AMS behavioral model; analog neural networks; current mode architecture; electrical simulation; incremental onchip learning; parallel weight perturbation algorithm; submicronics technologies; Algorithm design and analysis; Analog computers; Computer networks; Convergence; Electronic mail; Integrated circuit modeling; Mathematical model; Network-on-a-chip; Neural networks; Noise robustness; CAD; VHDL-AMS; WP algorithm; analog implementations; analog neural networks;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Industrial Electronics, 2004 IEEE International Symposium on
Print_ISBN :
0-7803-8304-4
Type :
conf
DOI :
10.1109/ISIE.2004.1571809
Filename :
1571809
Link To Document :
بازگشت