Make Deep Learning Great Again: Character-Level RNN Speech Generation in the Style of Donald Trump

A character-level recurrent neural network (RNN) is a statistical language model capable of producing text that superficially resembles a training corpus. The efficacy of these networks in mimicking Shakespeare, Linux source code, and other forms of text have already been demonstrated. In this paper, we show that character-level RNNs are capable of very believably mimicking the language of President Donald J. Trump after training on a corpus of speech transcriptions. We believe our most significant contributions to the study of character-level statistical language models are in sampling methodologies; specifically, we propose that introducing dropout during the text-generation phase introduces randomness that leads to more believable text.