School of Computing

Miss Farhana Liza

Research Student

Photo of F Liza
  • Tel:     +44 (0)1227 82 3192
  • Fax:     +44 (0)1227 762811
  • Email: fl207@kent.ac.uk
  • Room S15
    School of Computing
    University of Kent, CT2 7NF

Publications

My publications are available from the University of Kent's Academic Repository.

Recent News

  • September 13 -- Invited talk on "Automatic detection of new 
word meanings from social media." at NLP reading group in The Alan Turing Institute, London, UK (slides).
  • June 2018 to September 2018 -- Turing Intern at The Alan Turing Institute, London, UK (Project Page).
  • June 2018 -- Won Sciences Postgraduate Research Prize for 2018 (News Story).
  • November 2017 -- Our paper on language modeling has been accepted to AAAI 2018.
  • October 2016 -- The slide from the research talk on Spectral learning for Hidden markov model (HMM) at Computational Intelligence Group Seminars can be downloaded from here

PhD Project Summary

In the PhD project, I am investigating the Language Models using Deep Learning Techniques. Language modelling can be seen as a very large scale multi-class classification problem. The Language model defines a function that classify a sequence of words (context) to a word that is most likely to follow seeing the given context. Language modelling can be seen as a sequence prediction problem, where given a sequence of words, a discriminative model needs to predict the next word.

Statistical Language Models have lacking in incorporating really long term dependencies in a given context (a sequence of words). Neural language model can approximate the large context statistical language model, however introduces time complexity.

In this project I am investigating approaches that can understand and capture the hierarchical representation at different steps in the sequence data and use that dense representation to improve the accuracy of Language Models in a time efficient manner.

Research Interests

I am a member of the following research groups:

My chief area of interest are deep learning, latent variable models, representation learning and language modelling.

Programme Committee Member

  • AAAI Conference on Artificial Intelligence (AAAI), 2019.
  • RepEval 2017: The Second Workshop on Evaluating Vector Space Representations for NLP, Co-located with EMNLP 2017.

Reviewer

  • International Conference on Learning Representations (ICLR), 2019.
  • Annual Conference on Neural Information Processing Systems (NIPS), 2018.
  • Special Issue of The Knowledge Engineering Review journal, 2018.

Teaching

  • CO584: Solving Problems with Data : Practical Sessions
  • CO520: Further Object Oriented Programming
  • CO320: Introduction to Object Oriented Programming
I also help undergraduate students with "Stage 1 Unix classes for project week".

Personal Website

For my personal website please click here

School of Computing, University of Kent, Canterbury, Kent, CT2 7NF

Enquiries: +44 (0)1227 824180 or contact us.

Last Updated: 24/09/2018