Curriculum Learning for Cross-Lingual Data-to-Text Generation With Noisy Data

TL;DR

This paper introduces a curriculum learning approach for cross-lingual data-to-text generation with noisy data, improving output quality and faithfulness across multiple languages by using alignment-based sample ordering and annealing schedules.

Contribution

It proposes novel curriculum criteria tailored for cross-lingual DTG with noisy data, demonstrating significant performance improvements over existing methods.

Findings

BLEU score increased by up to 4 points

Faithfulness and coverage improved by 5-15%

Effective across 11 Indian languages and English

Abstract

Curriculum learning has been used to improve the quality of text generation systems by ordering the training samples according to a particular schedule in various tasks. In the context of data-to-text generation (DTG), previous studies used various difficulty criteria to order the training samples for monolingual DTG. These criteria, however, do not generalize to the crosslingual variant of the problem and do not account for noisy data. We explore multiple criteria that can be used for improving the performance of cross-lingual DTG systems with noisy data using two curriculum schedules. Using the alignment score criterion for ordering samples and an annealing schedule to train the model, we show increase in BLEU score by up to 4 points, and improvements in faithfulness and coverage of generations by 5-15% on average across 11 Indian languages and English in 2 separate datasets. We make code and data publicly available