Home

Awesome

This repository contains code for our papers published in NeurIPS 2019 and NeurIPS 2020.

Quantum Embedding of the Knowledge (NeurIPS 2019)

A novel approach proposed (by Knowledge & Reasoning Group at IBM Research AI, India) to embed symbolic Knowledge Base (KB), inspired by the theory of Quantum Logic. The goal is to be able to replicate the logical structure of the KB and symbolic logical reasoning in the embedding space. We call this approach E2R (Embed2Reason).

This work is first accepted for publication at NeurIPS 2019 Conference with paper details as below:

Dinesh Garg, Shajith Ikbal, Santosh K Srivastava, Harit Vishwakarma, Hima Karanam, L Venkata Subramaniam, "Quantum Embedding of Knowledge for Reasoning", to appear in Proc. of Neural Information Processing Systems (NeurIPS) 2019, Vancouver, Canada, 2019.

Inductive Quantum Embedding (NeurIPS 2020)

In this work, we extend our NeurIPS 2019 paper. The original QE idea is limited to the transductive (not inductive) setting. Moreover, the original QE scheme runs quite slow on real applications involving millions of entities. This work alleviates both of these key limitations. As an application, we show that one can achieve state-of-the-art performance on the well-known NLP task of fine-grained entity type classification by using the inductive QE approach.

This work is first accepted for publication at NeurIPS 2020 Conference with paper details as below:

Santosh Srivastava, Dinesh Khandelwal, Dhiraj madan, Dinesh Garg, Hima Karanam, L Venkat Subramaniam, "Inductive Quantum Embedding", to appear in Proc. of Neural Information Processing Systems (NeurIPS) 2020, Virtual-only Conference.

The purpose of this repository is to open-source material relavant to our Quantum embedding work (under Apache 2.0 License). Details about the current set of folders are given below: