Augmenting BERT-style Models with Predictive Coding to Improve Discourse-level Representations
This addresses the challenge of discourse-level understanding in NLP, offering a domain-specific improvement for tasks like discourse analysis.
The paper tackled the problem of generating discourse-level representations in language models by augmenting BERT-style models with predictive coding, resulting in improved performance in 6 out of 11 discourse-related tasks, particularly in discourse relationship detection.
Current language models are usually trained using a self-supervised scheme, where the main focus is learning representations at the word or sentence level. However, there has been limited progress in generating useful discourse-level representations. In this work, we propose to use ideas from predictive coding theory to augment BERT-style language models with a mechanism that allows them to learn suitable discourse-level representations. As a result, our proposed approach is able to predict future sentences using explicit top-down connections that operate at the intermediate layers of the network. By experimenting with benchmarks designed to evaluate discourse-related knowledge using pre-trained sentence representations, we demonstrate that our approach improves performance in 6 out of 11 tasks by excelling in discourse relationship detection.