Continuous N-gram Representations for Authorship Attribution.

This paper presents work on using continuousrepresentations for authorship attribution.In contrast to previous work,which uses discrete feature representations,our model learns continuous representationsfor n-gram features via a neuralnetwork jointly with the classificationlayer. Experimental results demonstratethat the proposed model outperforms thestate-of-the-art on two datasets, while producingcomparable results on the remainingtwo.