Convolutional Neural Networks for Sentence Classification

The paper reports a series of experiments with convolutional neural networks trained on top of pre-trained word vectors for sentence-level classification tasks. A simple CNN with little hyperparameter tuning and static vectors achieves excellent results on multiple benchmarks, and the author additionally proposes a simple modification to the architecture that allows the use of both task-specific and static vectors.

Learning task-specific vectors through fine-tuning offers further gains in performance. The CNN models improve upon the state of the art on 4 out of 7 tasks, which include sentiment analysis and question classification, demonstrating that simple convolutional architectures over pre-trained word vectors are strong sentence classifiers.