The world's expertise is captured in layer upon layer of unstructured text. Compliance documents, legal texts, financial records, even web pages - there is a wealth of information available which our most advanced systems have difficulty comprehending.
Bloomsbury AI aims to unlock this expertise and enable its seamless communication by combining Machine Reading (MR), Deep Learning and Data Augmentation to provide experts with AI-driven, guided systems allowing them to transfer and share their knowledge more accurately and efficiently than ever before.
We present some insights from our research into transfer learning for open-domain question answering, together with a demonstration of the technology and a brief discussion of some ideas for further work.
YOU MAY ALSO LIKE:
- Infiniteconf 2019 - The conference on Big Data and AI (in London on 4th - 5th July 2019)
- IWDS 23: Intro to Natural Language Processing (NLP) - 1/2 (in London on 24th June 2019)
- Hate Speech as a Service (SkillsCast recorded in May 2019)
- From input to information – ML to the rescue! (SkillsCast recorded in March 2019)
Question Answering at Scale
Originally from a mechanical engineering background, Max followed up on his passion for machine learning and AI with a Masters from UCL. He now works on solving challenging problems focused on making the world's expertise more accessible.