University of Malta

UOM Main Page
Apply - Admissions 2016



Previous - 1 2 3 4 5 6 7 - Next

Can a Computer Perceive 3D Form from a Hand-drawn Sketch?

30 Oct 2009

The University of Malta participated in the DA EXPO which was held at the Intercontinental Hotel between 4 and 6 November 2009. As part of this participation, the Department of Systems and Control Engineering and the Department of Industrial Manufacturing Engineering  presented a 60-minute seminar entitled:

Can a Computer Perceive 3D Form from a Hand-drawn Sketch?

This seminar took place on the 4th November at 16:30. The speakers, Professor Kenneth Camilleri, Ms. Alexandra Bonnici and Dr. Philip Farrugia, threw light on the problems that machines have to handle to interpret the 3D form intended in a designer's hand-drawn sketch, an ability which humans take for granted because we tend to do it so well. The seminar disclosed the R&D work carried out by this team over a period of more than eight years and included videos demonstrating applications of sketch interpretation computer tools and a live demonstration illustrating one practical solution to this problem.

Talks on Spatio-temporal Modelling

22 Oct 2009

Date: Monday 26th October 2009
Time: 1500-1630
Location: Engineering Building Lecture Room EB3

The Department of Systems and Control Engineering is holding a seminar session consisting of two  presentations on modelling of spatio-temporal systems from measured data. Some practical aspects regarding data collection for spatio-temporal modelling will be considered, together with two different modelling and estimation methodologies.

Presentation 1:
Speaker: Mr. Andrew Zammit Mangion (PhD student, University of Sheffield, U.K.)
Title: Parameter Identification of spatiotemporal systems governed by SPDEs

Abstract: The modeling and identification of spatiotemporal (ST) systems is today an active area of research in various disciplines with typical frameworks including the use of partial differential equations (PDEs), integro-difference equations, and coupled map lattices, amongst others. However, few researchers in this field have taken up the task of using stochastic PDEs (SPDEs) as a tool for data-driven system identification. The theory associated with this class of models is relatively recent and much of the work developed so far is presented solely from a mathematical perspective, rather than from an engineer’s point of view.
The aim of this talk if is to propose identification techniques for this class of models which are easily accessible to the engineering community and which adhere to the traditional way in which these systems are studied. We present a rigorous means for breaking down these complex ST representations into more manageable stochastic state-space models which allow a whole range of standard tools to be applied to this relatively new class of application problems. Expectation maximisation (EM) together with a few extensions (supplemented EM and variational Bayes EM) are outlined as methods of choice for the identification procedure.

The talk concludes by demonstrating the application of the proposed techniques to the identification of heterogeneous systems such as a distributed parameter system governed by the stochastic diffusion equation. Both static and mobile sensing strategies are used and briefly discussed.

Presentation 2:
Speaker: Mr. Kenneth Scerri (Assistant Lecturer, University of Malta)
Title: Spatial Sampling and Reconstruction of Spatio-Temporal Systems

Abstract: Natural phenomena ranging from the spread of bacteria to the development of tropical storms exhibit complex spatio-temporal interactions. Central to the analysis, prediction and control of such systems is the development of accurate models to represent their behaviour. For systems where the scientific laws governing the system's behaviour are not well understood or too complex, models are inferred from gathered observations with the aim to replicate the important spatio-temporal patterns in the data.
While systems theory provides a wealth of knowledge for dealing with complex dynamic behaviour, its models and methods have rarely been utilised in a spatio-temporal domain. Thus the work presented in this talk aims to bridge the gap between spatio-temporal modelling and systems theory by first providing a novel state space representation for spatio-temporal behaviour and second developing methods to infer these models from data.
Data is usually gathered by localized sensors obtaining spatially discrete observations which are inevitably corrupted by measurement inaccuracies. Nevertheless, the underlying phenomena usually evolve on a continuous spatial domain. Thus, a method is presented to optimally reconstruct continuous dynamic spatial processes from noise corrupted observations. Assuming knowledge of the spatio-temporal interactions involved, this method makes use of a novel spatio-temporal state space model able to represent both the system's dynamics, based on the stochastic integro-difference equation and also the data gathering processes based on the sensor characteristics. If the spatio-temporal behaviour of the system is not known, probabilistic methods are developed to infer from the noise-corrupted data a model representing the spatio-temporal interactions governing the system's behaviour and also the spatially continuous dynamic process being studied. 

Talk on Blind Source Separation

25 May 2009

On the 2nd June 2009 the Department hosted Ms. Charmaine Demanuele, a PhD student at the University of Southampton U.K, who delivered a presentation on the Analysis of Very Low Frequency Neuronal Oscillations in Electromagnetic Brain Signal Recordings: A Blind Source Separation Approach.


On the Analysis of Very Low Frequency Neuronal Oscillations in Electromagnetic Brain Signal Recordings: A Blind Source Separation Approach

Charmaine Demanuele and Christopher J. James
Signal Processing and Control Group, ISVR
University of Southampton, Southampton, United Kingdom

Electrophysiology is a valuable tool for the diagnosis and prognosis of neuronal disorders as well as for providing general insight into human brain function. However, electro- and magneto- encephalographic (EEG/MEG) data is obtained from an inherently noisy recording process and typically contains a mixture of physiological and ambient artifacts, along with active brain sources. For this reason, it is required to efficiently isolate neurophysiologically meaningful sources from the recorded signals.

Independent component analysis (ICA), a blind source separation (BSS) technique for the extraction of statistically independent components from a set of measured signals, has been extensively used to achieve the separation of spatially distinct brain sources. Single channel ICA (SC-ICA) is a variant of ICA which can be applied to extract underlying processes from a single recording channel by using only temporal information inherent in the signal dynamics. Expansion of this technique has led to the development of Spacetime ICA (ST-ICA), whereby SC-ICA is applied to a number of recording channels, hence providing both temporal and spatial information to inform the standard ICA algorithm. This presents a specific form of BSS that exploits the rich, dynamical time structure of electrophysiological data as well as the multi-channel nature of the recordings.

This talk explores the use of these techniques for the analysis of spontaneous neuronal very low frequency oscillations (VLFO, <0.5Hz), previously regarded as ‘physiological noise’, in MEG data. These VLFOs, which appear to be intrinsically generated by the brain and occur within widely distributed neuroanatomical systems, have been consistently reported and analysed in blood oxygen level dependent (BOLD) imaging studies. They are thought to arise from variations in metabolic demands in the resting brain and are unrelated to cardiac and respiratory events. However, they also persist during tasks where they contribute to inter-trial variability in evoked responses and present a potential source of attention deficit during task performance. These novel BSS algorithms are effectively applied to the analysis of systems with high spatio-temporal complexity, and highlight the effect of the neuronal VLFOs on task-processing in normal and clinical groups (particularly children with Attention Deficit Hyperactivity Disorder).

Talks on Biomedical Signal Processing

12 Apr 2009

The Department of Systems and Control Engineering  hosted Dr. Bart Vanrumste from the Bioseciences and Technology Department of K.H. Kempen University College in Geel, Belgium between the 21st and 22nd April 2009.

Dr. Vanrumste delivered a series of three talks on Signal Processing for Biomedical Engineering. The topics of the talks were:

  • Detection of movement in patients with epilepsy during sleep.
  • EEG source localization.
  • A video system for the detection of pain in demented elderly.

New Website Launched

16 May 2008

Welcome to the new Department of Systems & Control Engineering website.

Previous - 1 2 3 4 5 6 7 - Next


Funding Award for BrainApp
CBC awarded research funding under the FUSION Technology Development Programme (TDP) 2017

Closing date for MSc course applications


The deadline for late applications for the MSc in Signals, Systems and Control is the 31st August 2017. Click here for more information.

TVM news feature


News feature on the work currently being carried out by our PhD student Ms Rachael Nicole Darmanin, supervised by Dr Ing. Marvin Bugeja. Link here.

PhD scholarships at University of Sheffield, UK


Please click here for more information.

Postdoctoral Position in Robotics and Control


Postdoctoral Position at the University of Le Havre in France. Click here for more information.

MSc by Research - Call for Expression of Interest


The Department of Systems and Control Engineering and the Centre for Biomedical Cybernetics are seeking students who are interested in pursuing studies leading to an MSc in Engineering degree (mainly by research) on any of the topics detailed here.

Postdoctoral Research Position in EEG


[CLOSED] Postdoctoral Research Position in EEG Signal Processing with the Centre for Bionedical Cybernetics. Click here for more information.


Log In back to UoM Homepage