Text Classification Improved by Integrating Bidirectional LSTM with Two-dimensional Max Pooling

TL;DR

This paper introduces a novel text classification model that integrates bidirectional LSTM with two-dimensional max pooling and convolution to better capture features across both sequence and feature dimensions, improving performance.

Contribution

The paper proposes applying 2D max pooling and 2D convolution to RNN-based text representations, enhancing feature extraction over traditional 1D pooling methods.

Findings

Achieves state-of-the-art results on Stanford Sentiment Treebank tasks.

Outperforms existing models on 4 of 6 benchmark datasets.

Demonstrates the effectiveness of 2D pooling in sequence modeling.

Abstract

Recurrent Neural Network (RNN) is one of the most popular architectures used in Natural Language Processsing (NLP) tasks because its recurrent structure is very suitable to process variable-length text. RNN can utilize distributed representations of words by first converting the tokens comprising each text into vectors, which form a matrix. And this matrix includes two dimensions: the time-step dimension and the feature vector dimension. Then most existing models usually utilize one-dimensional (1D) max pooling operation or attention-based operation only on the time-step dimension to obtain a fixed-length vector. However, the features on the feature vector dimension are not mutually independent, and simply applying 1D pooling operation over the time-step dimension independently may destroy the structure of the feature representation. On the other hand, applying two-dimensional (2D) pooling operation over the two dimensions may sample more meaningful features for sequence modeling tasks. To integrate the features on both dimensions of the matrix, this paper explores applying 2D max pooling operation to obtain a fixed-length representation of the text. This paper also utilizes 2D convolution to sample more meaningful information of the matrix. Experiments are conducted on six text classification tasks, including sentiment analysis, question classification, subjectivity classification and newsgroup classification. Compared with the state-of-the-art models, the proposed models achieve excellent performance on 4 out of 6 tasks. Specifically, one of the proposed models achieves highest accuracy on Stanford Sentiment Treebank binary classification and fine-grained classification tasks.