Optimal Transport Distillation for Cross-Lingual Retrieval in Low-Resource Languages
Description
Benefiting from transformer-based pre-trained language models, neural ranking models have made significant progress. More recently, the advent of multilingual pre-trained language models provides great support for designing neural cross-lingual retrieval models. However, due to unbalanced pre-training data in different languages, multilingual language models have already shown a performance gap between high and low-resource languages in many downstream tasks. And cross-lingual retrieval models built on such pre-trained models can inherit language bias, leading to suboptimal result for low-reso
Research goal: How does optimal transport distillation compare to other knowledge distillation techniques (e.g., contrastive distillation, soft-label distillation) in improving cross-lingual retrieval accuracy for low-resource languages in benchmarks like MIRACL or XLSD?
Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.1/10.