Skip to content
On this page

Research

Currently I'm working on research in NLP, specifically in understanding how positional bias works in language models and how we can mitigate it. I'll be working under Prof. Duygu Ataman on this research project.

I have experience in research on time series forecasting, and am currently working with Mr. Suhwan Chung on further enhancing accuracies and computational efficiency when doing time series forecasting.

My current interests lie in exploring NLP and ML to enhance insights generation and analysis for unstructured data like text and documents.

Undergraduate Thesis (Final Year Project)

Feature Selection for Demand Forecating incorporating external covariates (Access my thesis here)

During my final year, I enrolled in a research project in collaboration with Becton Dickinson (BD), under the supervision of Prof. Jagath Rajapakse from NTU and Mr. Suhwan Chung from BD.

Key outcomes -

  • Improvement in accuracy compared to benchmarks in multiple time series instead of just one.
  • Utilised semantically related external variables as features.
  • Developed a new approach to feature selection that prevents the curse of dimensionality.

URECA

I enrolled in an invite-only research programme in NTU known as URECA. Here I researched about forecasting exchange rates using Restricted Boltzmann Machines.

Key outcomes -

  • Developed understanding into deep learning
  • Learnt Tensorflow and PyTorch for developing neural network routines