Jonathan Watkins

I currently teach numerical computing to PhD students and have a number of research interests.

Research Interests in Data

  • Tensor Networks and their application to machine learning
  • Text analysis with a focus on adapting PLSA
  • Data Analytics for Education

Tensor Networks

Application of Tensor Networks to machine learning problems has seen recent interest following a paper by Stoudenmire. My current interest is in trying to understand the feature maps and the approximation of the data tensor in the matrix product state.

Text Analysis

We are currently developing an understanding of text analysis. This was driven by a desire to classify assessment questions written by teachers where content tags were not provided or badly tagged. Modern techniques for text analysis do exist but we have recently found the simple PLSA approach is allowing us to gain new insights into the structure of the corpus and what it means to be classified as being from a class.

Data Analytics for Education

We have applied clustering to diagnostic testing and weekly assessment results to understand where grouping based on diagnostic test results alone are representitive of future performance. [Reference talk on diagnostic testing]

We have also recently analysed the effect of providing feedback to students on future performance.