Extreme Model Compression for On-device Natural Language Understanding
Kanthashree Mysore Sathyendra, Samridhi Choudhary, Leah Nicolich-Henkin
Abstract
In this paper, we propose and experiment with techniques for extreme compression of neural natural language understanding (NLU) models, making them suitable for execution on resource-constrained devices. We propose a task-aware, end-to-end compression approach that performs word-embedding compression jointly with NLU task learning. We show our results on a large-scale, commercial NLU system trained on a varied set of intents with huge vocabulary sizes. Our approach outperforms a range of baselines and achieves a compression rate of 97.4% with less than 3.7% degradation in predictive performance. Our analysis indicates that the signal from the downstream task is important for effective compression with minimal degradation in performance.
BibTeX
@inproceedings{mysore-sathyendra-etal-2020-extreme,
title = "Extreme Model Compression for On-device Natural Language Understanding",
author = "Mysore Sathyendra, Kanthashree and
Choudhary, Samridhi and
Nicolich-Henkin, Leah",
editor = "Clifton, Ann and
Napoles, Courtney",
booktitle = "Proceedings of the 28th International Conference on Computational Linguistics: Industry Track",
month = dec,
year = "2020",
address = "Online",
publisher = "International Committee on Computational Linguistics",
url = "https://aclanthology.org/2020.coling-industry.15/",
doi = "10.18653/v1/2020.coling-industry.15",
pages = "160--171"
}