Nota bene - The following archives contain research-standard, unsupported code, mostly comprising Matlab implementations of the methods from our publications.
[CODE] Demo code is provided to support the DeepAMR paper, in which deep learning models are developed for handling genomic data from tuberculosis bacteria, as described in Yang, et al. (2019).
Multilabel Learning for Tuberculosis Resistance Prediction
[CODE] This code archive supports Kouchaki et al., 2020 (in review).
Genomics of Tuberculosis
[CODE] This code was developed by Samaneh Kouchaki and Yang Yang, and relates to Yang et al. (2017) in Bioinformatics, in which resistance to antimicrobial drugs is predicted from genomic sequences from tuberculosis bacteria.
Condition Monitoring for Handpumps
[CODE] This code was developed for UNICEF, and describes various condition monitoring models for handpumps in the developing world. See, for example, Colchester et al. (2017).
Multivariate Extreme Value Statistics
[CODE] This Matlab code implements the various multivariate extensions to extreme value theory that we have published, the primary source for which is Clifton et al. (2010), with further information to be found in Hugueny et al. (2011). These journal articles pull together work from conferences, details for which may be found in , , and .
Generalised Pareto Distributions
[CODE] This code demonstrates the multivariate extension to the Generalised Pareto Distribution (GPD), from Clifton et al. (2011) and Clifton et al. (2013). The scripts assume that the Netlab toolbox is visible (for its sampling routines).
Extreme Function Theory
[CODE] These scripts implement the core equations from Clifton et al. (2013). These should run in a stand-alone manner - please see the README.txt for details.
NDtool - Novelty Detection Toolbox
[CODE] This toolbox implements methods described in Pimentel et al. (2014) and Clifton (2007).
MTGP - Multitask Gaussian Processes
[CODE] This toolbox provides an implementation of MTPGs for biosignal analysis.
[Toolbox] This Matlab toolbox implements many algorithms for estimating respiratory rate from cardiosynchronous signals, such as the ECG and PPG, and include "signal quality" algorithms.