Our experiments have confirmed the effectiveness of autoencoder models. Pay attention that we used fully connected neural layers to train autoencoders. Such models work with a fixed input data window. The algorithm we have built can training any models operating with a fixed input data window. But the architecture of recurrent models is different. To make a decision on the activation of neurons, such models also use their previous state, in addition to ...
Recent Visitors
Go to Profile
Mark as Read