Natural Language Processing

Lead Research Organisation: Imperial College London
Department Name: Computing

Abstract

Exploring reasoning capabilities in digital assistants, aptly titled, Reasoning in Digital Assistants. In this project I research the addition of reasoning capabilities
in the context of digital assistants in order to answer questions beyond basic information retrieval

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/N509486/1 30/09/2016 30/03/2022
1981060 Studentship EP/N509486/1 30/09/2017 30/03/2021 Nuri Cingillioglu
 
Description We have discovered that state-of-the-art deep learning techniques are able to learn how to perform logical reasoning. Formal reasoning consists of manipulation of symbols such as names Alice, Bob and relations between them mother, parent. Our approach started with no background knowledge and only through seeing examples of deductive reasoning, the neural network approach was able to learn multi-step inference.
Exploitation Route There is an active interest in creating deep learning systems that have foundations in formal reasoning such as logic. Thus, we have seen interest from private companies to collaborate with applications ranging from data-efficient reinforcement learning and query answering in noisy natural language sources.
Sectors Digital/Communication/Information Technologies (including Software)

 
Description We have approached by colleagues in DeepMind with the interest of incorporate and further researching the subject matter.
First Year Of Impact 2019
Sector Digital/Communication/Information Technologies (including Software)