Given a series of complex-valued training samples ðzi; yiÞ, i ¼ 1;2; . . . ; N, where zi 2 Cn and yi 2 Cm, the actual outputs of the single-hidden-layer feedforward network (SLFN) with complex activation function gcðzÞ for these N training data is given bywhere column vector wk 2 Cn is the complex input weight vector connecting the input layer neurons to the kth hidden neuron, bk ¼ ½bk1; bk2; . . . ; bkmT 2 Cm the complex output weight vector connecting the kth hidden neuron and the output neurons, and bk 2 C is the complex bias of the kth hidden neuron. wk zi denotes the inner product of column vectors wk and zi. gc is a fully complex activation function. The above N equations can be written compactly asand in practical applications the number N ~ of the hidden neurons is usually much less than the number N of training samples and HbaY, where
đang được dịch, vui lòng đợi..
