Analisis Fungsi Aktivasi Sigmoid Biner dan Sigmoid Bipolar dalam Algoritma Backpropagation pada Prediksi Kemampuan Siswa

Abstract

The backpropagation method is method which is very good to solve introduction complex designs of problem. In the backpropagation, network every unit in the hidden layer. The same thing is also hold to hidden layer. Every unit in the hidden layer is connected to every unit in out put layer. The activation function that use in backpropagation method is binary sigmoid function, bipolar sigmoid function. The characteristic that must be have by activation function is continue and not decrease static. From the result of research so can be collected the using of activation function biner sigmoid have the accuracy average level that is better than bipolar sigmoid. The speed of calculation is latter than bipolar sigmoid. The activation function of bipolar sigmoid is chosen depends of heuristic assumption which is stretch point (0, 1) that produce by the function of sigmoid is worse than stretch (-1, 1), that produce by function which bipolar sigmoid is not give influence to calculation point to neuron, compare with extreme point -1 which result by bipolar sigmoid function. This result is caused of extreme point 0 that produce by biner sigmoid function is give less influence to calculation point in neuron, compared with extreme point -1 which produce by bipolar sigmoid function.