Causally Denoise Word Embeddings Using Half-Sibling Regression -- AAAI 2020

Paper

Our paper "Causally Denoise Word Embeddings Using Half-Sibling Regression" is accepted by AAAI 2020. This is the link to the arXiv version (with appendix): https://arxiv.org/abs/1911.10524

This repository contains the code of the paper.

Requirements

• Python 3.6
• tensorflow 1.9
• pandas
• gdown
• requests