Coarse-Grain Parallelization of Neural Network-Based Face Detection Method
Volodymyr Turchenko, Ihor Paliy, Viktor Demchuk, Ruslan Smal, Lev Legostaev · 2007
The paper describes the coarse-graining paralleling of the neural networks' cascade training for the face detection method. The training parallelizing is done on Grid system using MPI technology. Experimental results show the decreasing of training time for the neural networks' cascade in 1.3 times compared with a sequential execution on a single uniprocessor machine.