Analysis, Normalization & Estimation Algorithms for the Stratification of AML/ALL Cancer Patients
October 2006 – Werner Van Belle
This is a postdoctoral proposal sent to the University in Bergen. This project aims to apply signal processing techniques to analyze, normalize and build models for cancer proteomics. Through linking biomedical parameters (such as survival rate) to specific protein samples (from patients), better therapy might come into reach. The project parallelly investigates data handling procedures for mass spectrometry (MALDI TOF), 2D electrophoretic gels and gene array data. The built models will be validated as estimators in a cancer research setting.
Where are the emotional cues in music ?
May 2006 – Werner Van Belle, Bruno Laeng
What are the underlying dimensions that give structure and meaning to music ? To answer this question, we aim to integrate methodological techniques from other disciplines (signal processing & mathematics) into the field of psychology and psycho-physiology. Our main goal is to measure in a mathematical and computational way musical parameters and relate those to the human judgment of a song its emotional content and in addition compare these to psycho-physiological measures (pupillometry and EEG).
Designing Music Therapy: Developing Algorithms to Extract Emotion from Music
November 2005 – Werner Van Belle, Bruno Laeng
The effect of music as a psychotherapeutic tool has been recognized for a long time. Alone, or in combination with classical treatment, music can alleviate depression, stress and anxiety, as well as acute and chronic pain. Such beneficial effects are likely to derive from its ability to induce mood changes. However, it remains unclear which aspects of music can cause emotional changes. This project aims to link advanced audio signal processing techniques to empirical psychoacoustic testing to develop algorithms that automatically retrieve emotions associated with a particular piece of music. Such algorithms could then be used to select and develop musical pieces for therapeutic purposes.
Denoising of MALDI-TOF, 2D Gels and NMR Spectra
June 2005 – Werner Van Belle, Kjell-Arild Høgda
In this project we aim to use signal processing techniques to remove noise and normalize data of three different spectroscopy techniques: MALDI-TOF mass spectrometry, 2D gel images and NMR spectroscopy. Removal of noise is important in order to obtain more accurate data, to allow for automatic analysis in high throughput proteomics and to understand experimental limitations. Normalization schemes are necessary to compare results between different machines and different samples. This research will lead to a better quantitative understanding of experimental inaccuracies and allow for quantified biological comparisons. The developed algorithms will be freely accessible through a web interface
Model Inference and Simulation of the PI3 Kinase Pathway
June 2005 – Werner Van Belle, Niels Aarsæther
Cell simulation is the mathematical modeling of the behavior of a cell such that the model can be algorithmically executed, thereby enabling visualized prediction of the behavior of real living cells. Such models will be useful in the future to optimize and regulate biological processes such as the growth of yeasts, fungi and to find new medicines which merely regulate and coerce the cell in working differently/correctly (cancer research, diabetes, …). In Norway, model inferencing techniques and learning techniques are being used at some institutions (Jonathan M. Irish, Randi Hovland et al, and Bø, T. H., Dysvik, B., Jonassen), nevertheless these models are seldom used in a predictive cell simulating context. As far as we know, there is currently no active research around whole cell simulation in Norway. From a methodological point of view it is clear that new methodological approaches that include molecular biology, cellular imaging, real-time kinetic analysis and network integrated analysis are required to progress in understanding the nature of signaling specificity. To keep track of and to quantify the complexity of pathways, a computational approach seems essential.
Development and Integration of Algorithms that Extract Emotion from Music
March 2005 – Werner Van Belle, Geir Davidsen, Bruno Laeng
The described project aims to link advanced audio signal processing techniques to empirical psychoacoustic testing in order to develop algorithms that can automatically retrieve (part of) the emotions humans associate with music. The project is set up as a cooperation between 3 partners: Norut IT, the Psychology department at the University of Tromsø and the music conservatory at Hoyskolen in Tromsø. Commercial relevance of the project is found in audio content extraction for databases and search engines, quantitative assessment of emotion in music for teaching, automatic creation of playlists, categorizing sound libraries and plugins for sound production software used in studios.
MarFlow Sample Tracking
November 2004 – Werner Van Belle
Tracking samples throughout many different organisations which all use different labeling protocols and are located at different geographical locations is difficult. This R&D program aims at the creation of a sample tracking system. The system will offer a) a labelling protocol designed for storage and unique identification of samples integrating many different labeling techniques b) a decentralised storage capacity (every organisation can store the data locally) and c) a security model which will take into account the ethics of sample exchange. A second aspect of the project aims at offering integration and data exchange towards existing sample tracking systems and analysis programs. We believe that this project and associated effort will reduce the costs of future information systems as well as increase cooperation between different research groups. The longer term goal of this project is the development of one or more commercial products related to sample tracking in cooperation with interested partners.
CodFormatics: Design of a Bioinformatic System to Assist the Cod Breeding Program
June 2004 – Tor Flå, Werner Van Belle, Said H. Ahmed, Madjid Delghandi
As input to the national codbreeding program, it is important to identify regions within the genome that are responsible for quantitative properties (QTL’s). Obtaining these however is complicated by the huge amount of involved (environmental) factors. To be able to approach this problem we propose the creation of a bioinformatics tool that will a) help in analysing current datasets and b) help in guiding future experiments. E.g; by carefully selecting a set of fishes with the most desirable properties it might be possible to find the QTL much quicker than what would be possible by analysing all fishes.
Trustnet: Scalable, Trusted Information Sharing for Ambient Applications
June 2004 – Werner Van Belle, Lars-Kristian Vognild, Tage Stabell-Kulø, Theo D’Hondt
Peer-to-peer systems are systems in which the gross of the data is directly communicated between hosts, without going through centrally placed servers. Typically, this kind of networks organize themselves automatically, thereby neglecting the actual network topology. This makes peer-to-peer systems a very attractive means to support ad-hoc networks, such as interconnected embedded devices. However, the peer-to-peer paradigm still poses a number of research problems. Among them scalability, finding useful information, security and the programming of applications in such volatile environments. The objective is to address these problems by developing a scalable and secure peer-to-peer information sharing platform together with a suitable programming model. The consortium consists of three partners. Norut IT, which is an applied research institute, the department of computer science at the University of Tromsø (UiTø), which pioneered the now wide spread distributed agent paradigm in 1996, and the Programming Technology Lab at Brussels University, which is doing research in programming language engineering for distributed systems. By coupling the experience of these partners we ensure a strong, well balanced consortium.
VRT MPG Project
May 2001 – Peter Schelkens, Bart Wouters, Werner Van Belle, Rudy Lauwereins, Jan Cornelis, Theo D’Hondt, Lode Nachtergaele, Kris Van Bruwaene, Harry Sorgeloos
The research project is part of a transition toward a content-management system. This content management system should be able to manage all new and old media. This includes images, sound, text, graphics, games and interactive scenario’s…). The development of a content management system implies a transition toward digital media. The MPG project is on one track responsible for doing this conversion, in such a way that releasing media on future communication channels (such as games via television) is facilitated (IMEC). On a second track all this media-content should be easily retrievable and automatically accessibly throughout whole the corporation. Therefore research is being conducted in component-based systems and ontologies (PROG). The total budget of this project is 2’478’940 EUR.
Aanvraag IWT Specialisatiebeurs: Positie Optimalisering binnen Mobiele Multi-Agent Systemen
September 1998 – Werner Van Belle, Wolfgang De Meuter, Theo D’Hondt
A brand new programming paradigm is that of mobile software agents. One is now able to move an agent from one machine to another while its execution state remains intact. It is clear that interaction between two agents on the same host is much faster than the same interaction between agents on different hosts. Because of this the performance of these systems can be finetuned by moving agents to the same location if they are interacting enough. (This is done of course while the system is running.) The only problem is the unpredictability of the interaction patterns between the agents. For this reason we can’t make these systems run globally better than open distributed systems. We can’t predict that agent A should always be near agent B in order to enhance performance. In this doctoral study we want to develop the necessary algorithms and methodologies to automate the distribution of mobile agents, with the main purpose of obtaining better global performance. The methodology we will use consists of moving agents towards each other. If agents interact on a local machine an after the interaction retract to their original position it is possible to lower the response time. This will be done by an additional management layer to existing mobile multi-agent systems. This extra will be able to reason about the agents’ positions at a different, higher level, while the agents carry out their tasks. On this level we will develop some models to describe the task execution of the agents (how they respond to certain messages, what their standard behavior is and so on). It is important to know that these models will only use local information. After the agent models we will develop algorithms to fulfill our Quest, that is, an automated higher global performance. To achieve this we will look at genetic programming, subsymbolic statistic techniques and learning techniques like reinforcement learning.
Object Gerichtheid en Message-Passing over Wide Area Networks – Mogelijke Agent-Strategien
January 1997 – Werner Van Belle, Wolfgang De Meuter, Patrick Steyaert, Theo D’Hondt