Multimodal Pre-training for Enhanced Zero-Shot Cross-Lingual Transfer in SALT

This paper studies zero-shot cross-lingual transfer of vision-language models. Specifically, we focus on multilingual text-to-video search and propose a Transformer-based model that learns contextualized multilingual multimodal embeddings. Under a zero-shot setting, we empirically demonstrate that performance degrades significantly when we query the multilingual text-video model with non-English sentences. To address this problem, we introduce a multilingual multimodal pre-training strategy, and collect a new multilingual instructional video dataset (MultiHowTo100M) for pre-training. Experimen

Research goal: Can SALT's zero-shot cross-lingual transfer performance be further improved by incorporating multimodal pre-training (e.g., using CLIP or OWL-ViT) on image-text pairs, as measured by F1 scores on MLQA and XCOCO?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.5/10.