@inproceedings{Zouhar_2022_Base,
title = {Knowledge Base Index Compression via Dimensionality and Precision Reduction},
author = {Vil{\'e}m Zouhar and Marius Mosbach and Miaoran Zhang and Dietrich Klakow},
url = {https://arxiv.org/abs/2204.02906},
year = {2022},
date = {2022},
publisher = {Spa-NLP workshop at ACL 2022},
address = {22nd-27th May 2022 Dublin, Ireland},
abstract = {Recently neural network based approaches to knowledge-intensive NLP tasks, such as question answering, started to rely heavily on the combination of neural retrievers and readers. Retrieval is typically performed over a large textual knowledge base (KB) which requires significant memory and compute resources, especially when scaled up. On HotpotQA we systematically investigate reducing the size of the KB index by means of dimensionality (sparse random projections, PCA, autoencoders) and numerical precision reduction.
Our results show that PCA is an easy solution that requires very little data and is only slightly worse than autoencoders, which are less stable. All methods are sensitive to pre- and post-processing and data should always be centered and normalized both before and after dimension reduction. Finally, we show that it is possible to combine PCA with using 1bit per dimension. Overall we achieve (1) 100<span id="MathJax-Element-1-Frame" class="MathJax" tabindex="0"><span id="MathJax-Span-1" class="math"><span id="MathJax-Span-2" class="mrow"><span id="MathJax-Span-3" class="mo">×</span></span></span></span> compression with 75%, and (2) 24<span id="MathJax-Element-2-Frame" class="MathJax" tabindex="0"><span id="MathJax-Span-4" class="math"><span id="MathJax-Span-5" class="mrow"><span id="MathJax-Span-6" class="mo">×</span></span></span></span> compression with 92% original retrieval performance.},
pubstate = {published},
type = {inproceedings}
}