Bayes Centre

Partner Event - Large scale structure aware anaphora resolution using QNLP

About the event

Mehrnoosh Sadrzadeh will give a talk, in person and online, for the Coffee House Tech Talk Series. 

 Natural language is structured into phrases, sentences, and pieces of text. One can derive statistical information from this data, and this has been the focus for many NLP projects, e.g. the ones leading to the GPT and BERT language models. Taking a complementary approach, we focus on statistics as well as  structure and show how this combination can be learnt using tensors. Learning tensors is hard for classical computers, a more natural approach is thus delegating the problem to quantum computers. In previous work, we extended the underlying logic of QNLP with modalities to be able to reason aout text and experimented with a small 140-entry anaphora resolution dataset. I will go over the extended logic and also present our most recent set of results, on a large 1,500-entry dataset. Our model outperformed a bag of word baseline, a discourse-only baseline, and a classical coreference resolution algorithm. This is joint work with my PhD students Lo Ian Kin, Lachlan McPhear,  Hadi Wazni.


Speaker Bio

I studied computer software engineering and then logic at the BSc and MSc levels at Sharif University of Technology, Tehran Iran. I got my PhD after moving from university of Ottawa to Quebec at Montreal in 2006, while being a visitor at Oxford 2003-2006.  Then I did two postdocs in Southampton and Paris, followed by two EPSRC fellowships (postdoctoral and Career Acceleration) and a Wolfson College Junior Research Fellowship at Oxford 2008-2012. In 2013, I got my first faculty position in QMUL. In 2019 I moved to UCL, where I became professor in 2022. At the moment I hold a Royal Academy of Engineering Research Chair.

Jun 20 2023 -

Partner Event - Large scale structure aware anaphora resolution using QNLP

Join Huawei and Mehrnoosh Sadrzadeh for the next tech talk, taking place in person and online on 20th of June.

Bayes Centre 4th floor, Bayes Coffee House

Meeting details:
Meeting ID:98933570