Title :
A digital signal processor based accelerator for artificial neural network simulations in a personal computer environment
Author :
Proctor, A. ; Hudson, W.B.
Author_Institution :
EECE Dept., Kansas State Univ., Manhattan, KS, USA
Abstract :
A method to accelerate artificial neural network calculations on personal computers for an educational environment is presented. The method utilizes the advanced architecture of digital signal processors (DSP) in the Texas Instruments (TI) TMS320 family to conduct network simulations. Network performance is greatly enhanced due to the efficiency with which the DSP´s accomplish the necessary computations relative to conventional processors. To further enhance performance a look-up table of precalculated transfer function values is used with the DSPs. The advantages of DSP with artificial neural network simulations are given. A detailed description of the design is presented
Keywords :
add-on boards; backpropagation; computational complexity; computer science education; coprocessors; digital signal processing chips; feedforward neural nets; microcomputer applications; neural net architecture; parallel architectures; table lookup; DSP; TMS320; Texas Instruments; artificial neural network simulations; digital signal processor; educational environment; look-up table; personal computer; transfer function; Acceleration; Artificial neural networks; Computational modeling; Computer architecture; Computer networks; Digital signal processing; Digital signal processors; Instruments; Microcomputers; Table lookup;
Conference_Titel :
Instrumentation and Measurement Technology Conference, 1993. IMTC/93. Conference Record., IEEE
Conference_Location :
Irvine, CA
Print_ISBN :
0-7803-1229-5
DOI :
10.1109/IMTC.1993.382567
