Jointly Trained Sequential Labeling and Classification by Sparse Attention Neural Networks

TL;DR

This paper introduces a jointly trained LSTM-based model with sparse attention for simultaneous sentence classification and sequence labeling, leveraging task correlations to improve performance in language understanding tasks.

Contribution

The paper proposes a novel joint training framework with sparse attention mechanism for concurrent sentence classification and sequence labeling using LSTM networks.

Findings

Outperforms baseline models on ATIS and TREC datasets

Effectively leverages task correlations for improved accuracy

Introduces a novel sparse attention mechanism

Abstract

Sentence-level classification and sequential labeling are two fundamental tasks in language understanding. While these two tasks are usually modeled separately, in reality, they are often correlated, for example in intent classification and slot filling, or in topic classification and named-entity recognition. In order to utilize the potential benefits from their correlations, we propose a jointly trained model for learning the two tasks simultaneously via Long Short-Term Memory (LSTM) networks. This model predicts the sentence-level category and the word-level label sequence from the stepwise output hidden representations of LSTM. We also introduce a novel mechanism of "sparse attention" to weigh words differently based on their semantic relevance to sentence-level classification. The proposed method outperforms baseline models on ATIS and TREC datasets.