Published Paper Details:

Cacography Detection is having appeal in business and scholastics. As of late loads of good work has been done on transcribed digit acknowledgment to improve precision. Manually written digit acknowledgment framework needs bigger dataset and long preparing time to improve exactness and diminish mistake rate. Preparing of Neural Organizations for huge informational collections is tedious undertaking on central processor. Henceforth, in this paper we introduced quick effective counterfeit neural organization for manually written digit acknowledgment on GPU to decrease preparing time. Standard back proliferation (BP) learning calculation with multi-facet perceptron (MLP) characterization is picked for this undertaking and executed on GPU for equal preparing. This paper zeroed in on explicit parallelization climate Process Bound together Gadget Engineering (CUDA) on a GPU consequently adequately speedup preparing and diminish preparing time.