Representation Theory of Recurrent Neural Network

In this paper, we use the representation morphism concept to analyze the connection between two recurrent neural networks, primarily when we evaluate the neural network function between two isomorphic neural networks. We construct the set of all isomorphic classes of recurrent neural networks. We build the set by the action of the isomorphism group on the set of all recurrent neural networks that have invertible weight. By the group’s action, we get the set of orbits and call it the moduli space. We analyze the moduli space to get its dimensions.