LiteMuL: A Lightweight On-Device Sequence Tagger using Multi-task Learning

TL;DR

LiteMuL is a compact, multi-task learning-based sequence tagger designed for on-device NLP tasks like NER and POS tagging, achieving high accuracy with minimal memory usage on mobile devices.

Contribution

This paper introduces the first on-device multi-task neural model for sequence tagging, significantly reducing model size while improving accuracy over existing methods.

Findings

Model size is approximately 2.39 MB.

Achieved 94.33% NER accuracy and 90.90% POS accuracy on CoNLL 2003.

Outperforms state-of-the-art models with 50-56% smaller size.

Abstract

Named entity detection and Parts-of-speech tagging are the key tasks for many NLP applications. Although the current state of the art methods achieved near perfection for long, formal, structured text there are hindrances in deploying these models on memory-constrained devices such as mobile phones. Furthermore, the performance of these models is degraded when they encounter short, informal, and casual conversations. To overcome these difficulties, we present LiteMuL - a lightweight on-device sequence tagger that can efficiently process the user conversations using a Multi-Task Learning (MTL) approach. To the best of our knowledge, the proposed model is the first on-device MTL neural model for sequence tagging. Our LiteMuL model is about 2.39 MB in size and achieved an accuracy of 0.9433 (for NER), 0.9090 (for POS) on the CoNLL 2003 dataset. The proposed LiteMuL not only outperforms the current state of the art results but also surpasses the results of our proposed on-device task-specific models, with accuracy gains of up to 11% and model-size reduction by 50%-56%. Our model is competitive with other MTL approaches for NER and POS tasks while outshines them with a low memory footprint. We also evaluated our model on custom-curated user conversations and observed impressive results.