Prediction of Protein Domain-Types by Backpropagation

摘要: We propose a neural net solution for the recognition of do main-types or pmLeins, which is H. IlH.f'd and irnporLanL problem in hiol­ og)'. WI:' 1m'll:' rOlJnd thaL w;ing clever· pf'f'pron�ssif1g ('edHliqIJp. r'elaLively HTlall nC II ral ncLwork..; perrorrn HII rpT·iscflgl.y well. Thc perrorrrmTices nets were measured by cross-validation Hoeffding's inequality ,yas utilized estimation confidence interval estimates.