Profile icon

Yuan Shen

Independent Research Fellow

Computing and Technology

Role

Yuan Shen is a Independent Research Fellow in the School of Science & Technology.

Career overview

Yuan Shen obtained his diploma and PhD degree in Physics at University of Zurich, Switzerland. After his PhD study, he moved on to receive research training in the area of Computational Statistics / Machine Learning at University of Warwick, Aston University, and University of Birmingham. In 2017, He took up a lecturer position at Department of Mathematical Science of Xi'an Jiaotong - Liverpool University (XJTLU). In 2018, he left XJTLU for an Independent Research Fellow position at NTU.

Research areas

  1. Development of MACHINE LEARNING techniques for statistical SIGNAL PROCESSING (MLSP)
    1. Bayesian inference in dynamical systems. Examples:
      1. Variational Markov Chain Monte Carlo for Diffusion Models
      2. Variational Gaussian Process Approximation for Diffusion Models
    2. Learning in model space. Examples:
      1. Classification of sparsely and irregularly sampled time series
      2. Multi-subject modelling in model space
    3. Probabilistic modelling. Examples:
      1. Modelling of fractal time series data
      2. Modelling of behavioural time series data
      3. Spatial modelling for multi-spectral retinal imaging data
      4. Spatio-temporal modelling of brain imaging data
    4. Incorporation of machine learning models. Examples:
      1. Gaussian Process based Surrogate Model for Bayesian Inverse Problem
      2. Variational Inference in Gaussian Process of Model Error for Model-order Reduction
  2. Incorporation of MLSP techniques for BIOMEDICAL INFORMATICS (BMI)
    1. State-of-the-arts learning paradigms for medical informatics. Examples:
      1. Learning Using Privileged Information for Cost-effective Diagnosis of Mildly Cognitive Impairment
      2. Learning in Model Space for Medication Response Prediction of Attention Deficit Hyperactivity Disorder
      3. Deep Learning for Time Series Analysis of sleep and epilepsy EEG
    2. Big Data Techniques for Feature Selection Topics in Medical Informatics Applications
      1. Big Data Techniques: (i) deep neural network model and (ii) higher-order tensor decomposition
      2. Feature Selection Topics: tba
      3. Medical Informatics Applications: tba

Research projects

  1. Markov Chain Monte Carlo for Models in Stochastic Geometry (an EPSRC project led by Dr Thonnes at The University of Warwick);
  2. Variational Inference in Stochastic Dynamic Models (an EPSRC project led by Dr Cornford at Aston University);
  3. Multispectral Retinal Image Analysis: A new technique for the assessment of Age-related Macular Degeneration (a Dunhill Medical Trust project led by Dr Styles at The University of Birmingham)
  4. Unified probabilistic modelling of adaptive spatial temporal structure in the human brain (a BBSRC project led by Prof Tino at The University of Birmingham);
  5. Personalised medicine through learning in the model space (an EPSRC project led by Prof Tino at The University of Birmingham).

Student supervision

  1. PhD student projects
    • Nahed Alowadi (completed in 2018) – Population-level spatio-temporal probabilistic modeling of fMRI data (Co-supervision with 25% capacity, Director of study is Prof Peter Tino at School of Computer Science of The University of Birmingham)
    • Hanin Alahmadi (completed in 2018) – Efficient Feature Extraction Methods for High-order Tensor Data (Co-supervision with 25% capacity, Director of study is Prof Peter Tino at School of Computer Science of The University of Birmingham);
  2. MSc/MRes student projects
    • Gulrukh Turabee (in progress) – Sleep stage classification with all-night  EEG recordings: a deep learning approach (Co-supervision with 50% capacity at Nottingham Trent University)
    • Ruchi Bharat Patel (in progress) – Unsupervised feature selection for gene scoring: a deep learning approach (Nottingham Trent Univeristy);
  3. FYP student projects
    • Shengjie Sun (completed in 2018) – Efficient Markov Chain Monte Carlo Algorithms for Bayesian Inference in Dynamical Systems (Supervision at XJTLU)