Fine-tuning Encoders for Improved Monolingual and Zero-shot Polylingual Neural Topic Modeling

TL;DR

This paper explores various fine-tuning strategies for transformer encoders to enhance neural topic modeling in monolingual and zero-shot polylingual contexts, demonstrating improved topic quality and cross-lingual transfer.

Contribution

It introduces multiple fine-tuning methods for encoders, including auxiliary tasks and integrated training, to improve neural topic models for multiple languages.

Findings

Fine-tuning on topic classification improves topic quality.

Integrating topic classification into training enhances cross-lingual transfer.

Any task-based fine-tuning significantly boosts model performance.

Abstract

Neural topic models can augment or replace bag-of-words inputs with the learned representations of deep pre-trained transformer-based word prediction models. One added benefit when using representations from multilingual models is that they facilitate zero-shot polylingual topic modeling. However, while it has been widely observed that pre-trained embeddings should be fine-tuned to a given task, it is not immediately clear what supervision should look like for an unsupervised task such as topic modeling. Thus, we propose several methods for fine-tuning encoders to improve both monolingual and zero-shot polylingual neural topic modeling. We consider fine-tuning on auxiliary tasks, constructing a new topic classification task, integrating the topic classification objective directly into topic model training, and continued pre-training. We find that fine-tuning encoder representations on topic classification and integrating the topic classification task directly into topic modeling improves topic quality, and that fine-tuning encoder representations on any task is the most important factor for facilitating cross-lingual transfer.