Radial basis function neural network approach for asymmetric cryptography

As technology evolves, artificial intelligence can be used in many aspects, including cryptography, such as artificial neural networks. One of the artificial neural network algorithms is the Radial Basis Function Neural Network (RBFNN). In a key generation, there are two artificial neural network architectures, namely arch1 and arch2. After obtaining the required values, the output of arch1 and arch2 is crossed into a pair of keys. The encryption process is carried out from the input to the hidden layer and the decryption process on the hidden layer to the output layer. The success of the system in performing encryption and decryption depends heavily on the center point. The number of centers with a maximum match percentage (100%) is the center point length from 30 to 90 characters with a width range of 0.698 -0.936. RBF does not affect the amount of input data. However, RBF is very influential on the input range during training because the RBF must pass all the data for mapping as shown by the interpolation theory