Matthew Gaddes

Matthew Gaddes

Profile

I am a PhD student working with Prof. Andy Hooper to detect signs of volcanic unrest in InSAR time series acquired by the Sentinel-1 satellites. We aim to achieve this through the application of several machine learning methods, and consequently during the course of my Phd I have become familiar with:

  • Blind signal separation methods:
    • Principal component analysis (PCA).
    • Spatial and temporal independent component analysis (ICA).
    • The ICASO algorithm for assessing the reliability of ICA results.
    • Non-negative matrix factorisation (NMF).
  • Manifold learning methods:
    • Multidimensional scaling (MDS).
    • t-distributed stochastic neighbour embedding (t-SNE).
  • Clustering methods:
    • Agglomerative clustering.
    • Density-based spatial clustering of applications with noise (DBSCAN).
    • Hierarchical DBSCAN (HDBSCAN).
  • Neural networks:
    • Use of Keras with Tensorflow.
    • Training of convolutional neural networks (CNN).

Teaching:

  • Computational inverse theory (SOEE3250)

Education History:

  • PhD Candidate in Tectonics, 2014 - present
  • M Earth Sci., The University of Oxford, 2006-2010

Supervisors:

I am funded by the 'Looking inside the continents from space' (LiCS) project.

Qualifications

  • <ul> <li>PhD Candidate in Tectonics, 2014 - present</li> <li>M Earth Sci., The University of Oxford, 2006-2010</li> </ul>