Research Associate in Deep Generative Models for Anomaly Detection

A research associate position for 30 months is available on the University Defence Research Collaboration (UDRC) phase 3 project jointly funded by EPSRC and UK Ministry of Defence. This position will contribute to the UDRC3 under work package 3.1, and will work within the Institute for Digital Communications at the University of Edinburgh.

This position provides a unique opportunity to work on Robust Generative Neural Networks, and develop novel techniques for anomaly detection on the basis of generative deep learning models.

The aim is to design architectures for generative models, e.g., Generative Adversarial Networks (GANs) or Variational Autoencoders, that are interpretable and operate on multi-modal data. GANs, for example, are state-of-the-art generative models that are trained by playing a game between two neural networks: one generating data that resembles the input (real) training data, and another discriminating between artificial and real data. Such a game enables learning the underlying distribution of the training data. Subsequently, the developed architectures will be applied to detect anomalies in data arising from several modalities. Challenges that the researcher will address include how to incorporate prior information into the design of the architectures and methodologies, make the learning process robust to the outliers, and how to deal with time-varying distributions.

This post is full time and fixed term for 30 months.

Salary: £33,199 - £39,609 per annum.

Closing Date: Thursday 18th October 2018 at 5pm (GMT).

UoE Vacancy Ref: 


Closing Date: 

Thursday, October 18, 2018

Contact Name: 

Contact Email: 

Contact Telephone: 

+44(0)131 650 5706