GLoT: A Novel Gated-Logarithmic Transformer for Efficient Sign Language Translation

TL;DR

This paper introduces GLoT, a Gated-Logarithmic Transformer designed to improve Sign Language Machine Translation by better capturing long-term dependencies, outperforming existing Transformer models in translation quality.

Contribution

The paper presents a novel GLoT model that enhances SLMT by effectively modeling long-term dependencies, addressing limitations of existing Transformer-based approaches.

Findings

GLoT outperforms baseline models across all evaluation metrics.

GLoT effectively captures long-term temporal dependencies in sign language data.

The approach shows promise for improving communication for the Deaf community.

Abstract

Machine Translation has played a critical role in reducing language barriers, but its adaptation for Sign Language Machine Translation (SLMT) has been less explored. Existing works on SLMT mostly use the Transformer neural network which exhibits low performance due to the dynamic nature of the sign language. In this paper, we propose a novel Gated-Logarithmic Transformer (GLoT) that captures the long-term temporal dependencies of the sign language as a time-series data. We perform a comprehensive evaluation of GloT with the transformer and transformer-fusion models as a baseline, for Sign-to-Gloss-to-Text translation. Our results demonstrate that GLoT consistently outperforms the other models across all metrics. These findings underscore its potential to address the communication challenges faced by the Deaf and Hard of Hearing community.