Hindi as a Second Language: Improving Visually Grounded Speech with Semantically Similar Samples
This work addresses the challenge of training multilingual models with imbalanced data, which is incremental but relevant for low-resource language applications.
The paper tackles the problem of language imbalance in bilingual visually grounded speech models by leveraging a high-resource language to improve a low-resource one, resulting in the low-resource language outperforming monolingual and bilingual baselines in cross-modal retrieval tasks.
The objective of this work is to explore the learning of visually grounded speech models (VGS) from multilingual perspective. Bilingual VGS models are generally trained with an equal number of spoken captions from both languages. However, in reality, there can be an imbalance among the languages for the available spoken captions. Our key contribution in this work is to leverage the power of a high-resource language in a bilingual visually grounded speech model to improve the performance of a low-resource language. We introduce two methods to distill the knowledge of high-resource language into low-resource languages: (1) incorporating a strong pre-trained high-resource language encoder and (2) using semantically similar spoken captions. Our experiments show that combining these two approaches effectively enables the low-resource language to surpass the performances of monolingual and bilingual counterparts for cross-modal retrieval tasks.