Mark Ebden | University of Oxford (original) (raw)

Mark Ebden

Related Authors

Graduate Center of the City University of New York

Uploads

Papers by Mark Ebden

Livre: Mental speed and cognitive ability ROBERTS Stephen, EVERSON Richard.

Complete Patent Searching Database and Patent Data Analytics Services.

In many data-driven machine learning problems it is useful to consider the data as generated from... more In many data-driven machine learning problems it is useful to consider the data as generated from a set of unknown (latent) generators or sources. The observations we make are then taken to be related to these sources through some unknown functionaility. Furthermore, the (unknown) number of underlying latent sources may be different to the number of observations and hence issues of model complexity plague the analysis.

Abstract This paper presents methods of analysis of electroencephalogram (EEG) signals using arti... more Abstract This paper presents methods of analysis of electroencephalogram (EEG) signals using artificial neural networks, and subsequent methods of detection of obstructive sleep apnoea (OSA) from the neural network outputs. EEG signals are measurements of scalp potential differences arising from the brain's electrical activity. Gross changes in the human EEG occur between different types of sleep.

: We present a fully automatedapproach to eye-movement artefact removalwhich uses time-varying un... more : We present a fully automatedapproach to eye-movement artefact removalwhich uses time-varying unmixing matricesfrom an ICA algorithm. We specically codea model which avoids removal of EEG informationin cases where the EEG corrupts theEOGs. IntroductionDue to the small magnitude of the EEG signal (afew tens of V) contamination from movements ofeye movements may be signicant. Subtractive techniquesbased on linear decompositions of EEG andEOG signals have been popular for eye-movementartefact removal [4].

This review outlines the theory of spectral estimation techniques based on the fast Fourier trans... more This review outlines the theory of spectral estimation techniques based on the fast Fourier transform (FFT) and autoregressive (AR) model and their application to the analysis of human tremor data. Two FFT-based spectral estimation techniques are presented, the Blackman-Tukey and periodogram methods. Factors that influence the quality of spectral estimates are discussed including the choice of windowing function.

The detection of novel or abnormal input vectors is of importance in many monitoring tasks, such ... more The detection of novel or abnormal input vectors is of importance in many monitoring tasks, such as fault detection in complex systems and detection of abnormal patterns in medical diagnostics. We have developed a robust method for novelty detection, which aims to minimize the number of heuristically chosen thresholds in the novelty decision process. We achieve this by growing a gaussian mixture model to form a representation of a training set of “normal” system states.

Abstract During obstructive sleep apnea, transient arousal at the resumption of breathing is coin... more Abstract During obstructive sleep apnea, transient arousal at the resumption of breathing is coincident with a substantial rise in blood pressure. To assess the hemodynamic effect of arousal alone, 149 transient stimuli were administered to five normal subjects. Two electroencephalograms (EEG), an electrooculogram, a submental electromyogram (EMG), and beat-to-beat blood pressure (Finapres, Ohmeda) were recorded in all subjects.

Biophysics Biophysics Biomedical engineering Biochemical engineering Medicine Biophysics Biomedic... more Biophysics Biophysics Biomedical engineering Biochemical engineering Medicine Biophysics Biomedical engineering Medicine.

We describe a Variational Bayesian (VB) learning algorithm for 1-dimensionalmixture models. VB mo... more We describe a Variational Bayesian (VB) learning algorithm for 1-dimensionalmixture models. VB model order selection is compared to the Minimum DescriptionLength (MDL) criterion and the algorithm is demonstrated on a numberof synthetic data sets. 1 IntroductionAttias has recently [2] described aVariational Bayes framework for GraphicalModels'. The key property of a Graphical Model in this context is that ithas hidden variables.