Presentations

BpmDj = Tempo detection, rhythm extraction, nearest neighbors, bit-slicing & distance metric learning
March 2014 – presented at Phonetisches Laboratorium; Universität Zürich – Werner Van Belle

BpmDj is a tool to mix music on your phone. It uses beatgraphs to visualize the music in 2 dimensionsd. By sliding such images over each other, the user can align and mix his music. The software also helps the DJ to navigate his music collection by calculating the nearest neighbors towards any song in its database. The talk focuses on the scientific aspects of BpmDj. It starts off with fine tunings we made to an autodifference-based tempo detector. Then we explain how rhythmical patterns are extracted and normalised, which provides us with a series of content vectors in a >1000 dimensional space. To store these points we created a novel data structure that allows us to find the k nearest neighbors by accessing less than 30% of the data, and with 20% less storage than needed by vector approximation files. The last part of the talk focuses on a fast converging distance metric learning. It learns the weights of the various dimensions by using their cross correlation function. Thereby it learns a true distance metric (as opposed to a distance function).
Finding K Nearest neighbors: ~30% data access, >1000 dimensional uniform spaces.
August 2013 – presented at Observe, Hack, Make 2013: Oudkarpsel – Werner Van Belle

This talk describes a method I developed to find the k nearest neighbors in large dimensional spaces with less than 30% data access. The research was conducted between 2010 and 2013 but only 2 days before the slides were made I figured out that the “VA-Files” approach (invented in 1998 by Weber) is very similar.
Measuring Tempo, Extracting Rhythm, Finding similar Music
August 2013 – presented at Observe, Hack, Make 2013: Oudkarpsel – Werner Van Belle

This talk demonstrates BpmDj, its beatgraphs, how it finds similar music and how transitions can be edited. The talk goes into the details of tempo measurement and how autodifferencing can be improved to provide results that contain much less 3/5/7 harmonics. After explaining tempo measurement, we describe how feature vectors respresenting a musical rhythm can be extracted. These feature vectors are located in a multi dimensional space and of allow for nearest neighbor detection. However, in order to make that work one must have an appropriate weighing scheme present. At the end of the talk we describe the development of a fast converging weighing scheme to learn distance metrics.
Measuring Tempo, Extracting Rhythm & Composition
August 2011 – presented at Chaos Communication Camp 2011; Finowfurt – Werner Van Belle

This talk explains how the tempo of music can accurately be measured and how it can be used afterwards to extract rhythmical information and composition properties. These can further be used to find similar songs, generate synthetic rhythms and mix music. The peaks/valleys in autodifference/autocorrelation plots correspond potentially to the perceived tempo of music. Consequently, they can be used to measure the tempo of music. The problem with these techniques is that reported tempos might be off by a multiplication factor. This presentation discusses removal of such biases as to flatten out the autodifference/autocorrelation plots. Secondly, we discuss how the problem of tempo-harmonics (120 BPM reported as 90 BPM for instance) can be reduced from a multi-class classification problem to a binary classification problem. The resulting tempo information makes it further possible to extract rhythm and composition extraction from music.
How to measure Tempo ? Autodifference, Autocorrelation, Rhythm Extraction and Neural Networks
October 2010 – presented at Technische Universität Dortmund – Werner Van Belle

The peaks/valleys in autodifference/autocorrelation plots correspond to the potentially perceived tempo of music. Consequently, they can be used to measure the tempo of music. The problem with these techniques is a certain bias, which is unique for each song and without its removal, reported tempos might be off by a multiplication factor. In this presentation we discuss removal of such biases. Secondly, we discuss how the problem of tempo-harmonics (120 BPM reported as 90 BPM for instance) can be reduced from a multi-class classification problem to a binary classification problem. We also argue that the use of ‘beats per minutes’ is an artificial measure and should be superseded with the use of ‘measures per minute’, as it turns out to be a much more robust measure. In the second part of the talk, we will focus on a variety of techniques to extract rhythmical information and classify time-signatures. We do so because the combination of time-signature and tempo forms the key to solve the tempo-multiplication problem. Although from a meta-scientific point of view, we know that a neural network should be able to classify various time-signatures, the implementation of such network is still somewhat elusive because it is unclear how to present the data to the network. We discuss a variant of bayesian classification as well as a number of neural networks we tried. Lastsly, we discuss the potential problems of creating such classifiers if we take into account the potential amount of data one might need to train a neural network properly.
Active Message Queues
January 2010 – presented at Duale Hochschüle Baden-Württemberg (DHBW); Lörrach – Werner Van Belle

This presentation briefly describes a useable abstraction to deal with multiuple tasks. The abstraction seperates tasks in space, in time, syntactically and dissociates the scheduler from the task. Multiple tasks can be processed through one thread, or we can choose to run each task in its own thread. The result is a lockfree programming style.
The Explosion of Data Sizes in Biomedical Sciences and the Need for Proper Size Reduction and Visualization
June 2009 – presented at ExistenzGründer Seminar; Lörrach; Germany – Werner Van Belle

Data sizes in biomedical sciences grow larger each year; while a microarray experiment in 2003 produced maybe 20kilobytes of data, a 2D gel analysis in 2004 produced 140Mb data, an analysis of mice tracking produced 40 gigabytes in 2006 and a single deep sequencing experiment produces around 1.5 Tb. To then further correlate multiple deep sequencing tracks, one is stuck with 40 terrabytes. Although we very likely reached some form of limit to this exponential ghrowth, the current state is problematic for researchers with and without IT background. To deal with this problem data integration and visualization tools are becoming increasingly important.
Deep Sequencing and a reanalysis of human histone methylation and acetylation patterns.
April 2009 – presented at Brain Research Institute; University of Zürich; Zürich; Switzerland – Werner Van Belle

This talk introduces sequencing by synthesis (sometimes also called solexa sequencing, deep sequencing or digtal gene profiling). This technology has been booming the past few years because it can measure an entire genome expression profile at basepair resolution without predesigned probes. This makes it an excellent explorative tool and depending on the preprocessing of the sample, different information can be obtained. At the cost of a single micro-array one can measure methylation/acetylation patterns, genomic DNA, detect SNP’s or measure the transcriptome. Never before has a single experiment, performed in 5 days, yielded 1.5Tb of data and 28’800’000’000 bases. Analyzing this data is a major challenge and the first part of the presentation covers the primary analysis. I will present the acquisition of images, calculation and normalization of intensities, base calling, short fragment quality filtering and how one aligns 800’000’000 short fragments to a genome of choice. The second part of the presentation touches upon some analysis problems and will illustrate our current approach to analyze chip-seq patterns of around 41 humane histone methylations and acetylations tracks. We will also present a major, yet computationally unresolved, challenge in dealing with alignment profiles.
Is Science an Art ?
January 2009 – presented at ETH Zurich, Department of Biosystems Science and Engineering (D-BSSE), Switzerland – Werner Van Belle

This talk presents some common logical traps in scientific reasoning as well as common statistical traps. In general our intuition is neither good in formal logic, nor is it good in statistics, making it necessary for scientists to consciously think about their scientific method. To complement this view we go into some detail into the value of values and discuss the widespread abuse of log values as well as the technique of fitting the data to the model, which is obnviously the wrong scientific direction. Sadly enough this is often found in peak finders and none of my experience so far supports that peak finding is a valueable method, from any point of view.
Reanalysis of Human Histone Methylation and Acetylation Patterns
January 2009 – presented at Symposium on Next Generation Sequencing; University of Lausanne – Werner Van Belle

We present a novel technique to analyze correlations between multiple sequencing tracks. This leads to a number of unique views on methylation patterns and through correlation maximization can we aggregate the information in a coherent manner. As last part of the talk we quickly introduce a normalization scheme that might improve correlation measures by adding noise.
Reanalysis of Histone Methylation and Acetylation Patterns
December 2008 – presented at ETH Zurich, Department of Biosystems Science and Engineering, Research Seminars – Werner Van Belle

We present a novel technique to analyze coprrelation between multiple sequencing tracks. Our preliminary results show that repeat areas have a dramatic impact on potential correlation values as well that GC content has little influence in the overall result. Fragment extension to 160BP is a useful choice with this technology.
Reanalysis of Histone Methylation and Acetylation Patterns
December 2008 – presented at Friedrich Miescher Institute (FMI) Computational Biology Seminars – Werner Van Belle

We present a novel technique to analyze coprrelation between multiple sequencing tracks. We also show unique views on methylation patterns through correlation maximization.
Sequencing by synthesis: from samples to data patterns.
November 2008 – presented at Novartis; Quantitative Biology Meetings – Werner Van Belle

I present a relative new technology: sequencing by synthesis (sometimes also called solexa sequencing, deep sequencing or digtal gene profiling). This technology has been booming the past few years because it can measure an entire genome expression profile at basepair resolution without predesigned probes. This makes it an excellent explorative tool and depending on the preprocessing of the sample, different information can be obtained. At the cost of a single micro-array one can measure methylation/acetylation patterns, genomic DNA, detect SNP’s or measure the transcriptome. Never before has a single experiment, performed in 5 days, yielded 1.5Tb of data and 28’800’000’000 bases. Analyzing this data is a major challenge and the first part of the presentation covers the primary analysis. I will present the acquisitation of images, calculation and normalization of intensities, base calling, short fragment quality filtering and how one aligns 800’000’000 short fragments to a genome of choice. The second part of the presentation touches upon the problem of presenting this data to biologists and will illustrate our current approach where we translate a biological question to a signal processing question without resorting to arbitrary cutoff values. For this purpose, I will recycle some histone methylation and acetylation patterns that were previously published.
Primary Analysis: From Sample to Wig file in 8 Easy Steps
October 2008 – presented at 1st workshop on high throughput sequencing – Werner Van Belle

This presentation explains how data processing is performed in the Illumina pipeline. It also extends on two necessary tools: Eland2Exp and Eland2Wig
The Illumina GA II Sequencing System
October 2008 – presented at Department of Computer Science; ETH Zurich; Zurich – Werner Van Belle

In this talk I present the Illumina software pipeline 1.0. This pipeline analyzes output from the GA2 sequencing by synthesis analysis system and produces 4 major outputs: intensity files, sequences, alignments and various error reports. Next to the overall pipeline layout I present a number of usage scenarios and document how they are realized within the pipeline.
What happened to my data ? Screening a mutual stabilized reporter system.
August 2008 – presented at ETH Zurich, Department of Biosystems Science and Engineering, Research Seminars – Werner Van Belle

Undisclosed
GOToolBox: Functional Analysis of Gene Datasets based on Gene Ontology
August 2008 – presented at Literature Seminar; ETH Zurich; Department of Biosystems Science and Engineering – Werner Van Belle

The presentation covers the GOToolbox. This toolbox uses an interesting comb inatorial approach to decude whether specific gene ontology classes are enriched or depleted. The talk focuses on the intuitive understanding of the reasoning behind the paper.
Deep Sequencing Unit: Inputs, Outputs, File Sizes & Formats
July 2008 – presented at ETH Zurich in Basel – Werner Van Belle

This presentation covers the various inputs, outputs and temporary files as created byy the Illumina Genome Analyzer Software v0.3. For the IT people it has an overview of the different file sizes.
Deep Sequencing Unit: Information Infrastructure
April 2008 – presented at ETH Zurich in Basel – Werner Van Belle

This presentation covers the current setup of the Deep Sequencing Unit as well as the challenges we currently face. The presentation will be used as a starting point to discuss a possible involvement of the CISD.
BpmDj and some High Level Concepts for Data Flow programming
January 2008 – presented at PLEIAD laboratory of the Computer Science Department of the University of Chile (Faculty of Engineering) – Werner Van Belle

This presentation was given at Pleiad to demonstrate the concepts I was working with. It contains two parts. The first part is BpmDj and was posed as a challenge: ‘try to write this in a general purpose high level language’. The second part of the presentation took some problems from bioinformatics (E.g: an array of movies) and discusses a suitable actor model and presents a tree based memory model in which each object can be readonly. This makes it possible to make an efficient file system in which we can avoid duplicates, which can easily be distributed. The language itself is a functional language with mutator operators, which in generla makes it possible to use caching (memoizing) techniques.
Biological Networks and Noise Propagation
January 2008 – presented at ETH Zurich in Basel – Werner Van Belle

I will present a) intensity dependent analysis of microarrays. This is an analysis technique that takes into account amplification errors and reports up/down regulations together with their 95% confidence intervals. I will explain the idea, the apparent improvement in analysis (x35 more genes reported) and the validation of this technique. The second topic focuses on b) assessing which protein are likely altered under specific gene knock-outs/knock-ins. We have been doing this by integrating a protein interaction map into micro array measurements and then propagating an influence value throughout that network. Although very rudimentary, this has given great results so far. I will give two example using MK5 and FKRP. I will wrap up the talk by focusing on normalization, measurement errors, biological variations and noise in genetic pathways.
2D Gel Correlation Analysis and Perspectives on Systems Biology
October 2007 – presented at Friedrich Miescher Institute for Biomedical Research, Basel – Werner Van Belle

Two-dimensional gel electrophoresis is a powerful technique to discriminate post-translationally modified protein isoforms. The first part of our presentation will focus on a technique we developed to relate such 2D gel images to specific external conditions. We will present the different steps of the technique, its validation on simulated images and give examples of its use on the analysis of P53-isoforms in AML patients. The second part of the presentation focuses on a recent experiment on biological networks. Microarray & qPCR make it easy to measure protein/transcription levels in cell systems. This technology can be used to sample the direct and indirect changes induced by cell system alterations, such as constitutive activation of genes or interfering with gene transciption (siRNAs). To provide a better overview of the effects on such alterations, we integrated a microarray measurement into a human protein interaction map. This made it possible to rank and cluster the proteins according to their expected relation towards the original change. Aside from providing a valuable tool for our research, our approach also revealed a number of substantial limitations which we will discuss in detail.
Microarrays, Protein Interaction Maps and Elevated Mazes
September 2007 – presented at Animal Sciences Group, Wagingen Universiteit, Lelystad – Werner Van Belle

The presentation consists of 3 parts. In the first part I present a novel method for the analysis of Microarrays. In brief, the method will, based on a number of replicates, create an intensity dependent error model. This model is then used to report confidence intervals on up/down regulation between different biological samples. The second part covers our use of these micro-array results by integrating them into a protein interaction map. A learning algorithm will then assess which impact the specific cell alteration potentially has on the proteome. The third part ellaborates on our functional validation of one of the predictions made by the reinforcement learner: MK5 influences anxiety. We validated this by means of transgenic mice on an elevated plus maze and in the light-dark box test
Micro-Array Confidence Intervals: a more detailed look & Protein Interaction Map Integration & Visualization
April 2007 – presented at Medical Genetics, University Hospital Northern Norway, Tromso – Werner Van Belle

This presentation explains the rationale behind probability distributions through a couple of simple examples. Its application on microarray measurements is presented. The second part of the presentation covers the integration of such micro arrya measurements into protein interaction maps.
Close-to-the-experiment Data Analysis
March 2007 – presented at LabForum, University Hospital Tromso (UNN) – Werner Van Belle

Correlation Analysis 2D Gels
February 2007 – presented at the High Technology Center of Bergen – Werner Van Belle

Micro Array Error Analysis and Integration/Visualisation of a Protein Interaction Map
January 2007 – presented at the Department Virology, University of Tromso and LabForum, University of Tromso & University Hospital of Tromso – Werner Van Belle

Concurrency Adaptation using Learning State Machines
August 2006 – presented at the computer science department of the university of Aberdeen – Werner Van Belle

Elaborates on the mediation of concurrency strategy conflicts in open distributed systems using learning state machines and constraint logic programming over Petri-nets
Systems Biology: Bridging the gap between experiment and interpretation
March 2006 – presented at Norut IT, Tromso – Werner Van Belle, Kjell-Arild Høgda

Systems Biology: Bridging the gap between experiment and interpretation
February 2006 – presented at the Institute of Biology, University of Tromso – Werner Van Belle, Kjell-Arild Høgda

Systems Biology: Bridging the gap between experiment and interpretation
January 2006 – presented at the Institute of Molecular Biology, University of Tromso – Werner Van Belle, Kjell-Arild Høgda

Data Driven Programming: Change Nets
November 2005 – presented at Norut IT, Tromso – Werner Van Belle

Change-nets are an extension to Petri-nets that allow optimal execution mapping onto thread based, process based connection oriented systems.
On the need for robust techniques
September 2005 – presented at the Department of Economic Sciences, University Leuven – Werner Van Belle

Biological Systems: Bridging the gap between experiments and interpretation
September 2005 – presented at the Functional Proteomics Unit, Departement of Medical Protein research, Flemish Institute for Biotechnology (VIB), Gent – Werner Van Belle

Advanced Signal Processing Techniques in BpmDj
August 2005 – presented at the Insomnia Festival 2005 – Werner Van Belle

Avanserte teknikker for lydsignalprosessering som kan vaere til stotte for moderne lydteknikere vil bli presentert. Teknikkene gir mulighet til a male lydfargen (basert pa psykoakustiske skalaer og PCA-analyse), ekkoegenskaper og sangens tempo (spenner fra halvautomatisk til helautomatisk sanntids tellere for slag/minutt) – som gir mulighet til a trekke ut informasjon om bade rytme og lydsammensetning. Lydteknikere kan finne dette nyttig for a trekke ut typisk lydmonstre fra sanger og sammenligne dem med hverandre. Visualiseringsteknikker for sangers “energilandskap” basert pa bolgeanalyser som kan brukes som et visuelt fingeravtrykk for hele sanger vol og bli presentert. Denne informasjonen kan lagres og ved hjelp av grupperingsteknikker automatisk generere spillelister. En demonstrasjon av Automatic Mix Generation vil ogsa bli gjennomfort.
Mk5 Cell Cycle Analysis
August 2005 – presented at the Department Virology, Institute Molecular Biology, University Of Tromso – Werner Van Belle

Analysis of wild type P53 protein distribution in cancer cells
May 2005 – presented at the 6th Bergen Conference on Cancer Research 2005 (BCCR2005) section: proteomics in cancer – Werner Van Belle, Nina Ånensen, Oystein Bruserud, Bjørn Tore Gjertsen

BpmDj Meta-Data Extraction
April 2005 – presented at the Music Conservatorium, Hoyskolen Tromso – Werner Van Belle

The presentation covers all meta data extraction aspects in BpmDj. It starts with explaining different techniques to measure tempo (energy enveloppe spectra, autocorrelation, peak detectors, ray-shooting and realtime tempo extraction). Beatgraphs are introduced, spectrum measurement based on the psychoacoustic bark scale is covered as well as echo characteristic, rythm and composition extraction. SVD / PCA analysis is used to map a multidimensional spectrum space to 3 primary colors.
Denoising of Maldi-Tof Mass Spectra & denoising of 2D gels
April 2005 – presented at the Programming Technology Lab, Computer Science Department, Vrije Universiteit Brussel – Werner Van Belle

BpmDj Meta-Data Extraction
March 2005 – presented at the Psychology Department, University of Tromso – Werner Van Belle

The presentation covers all meta data extraction aspects in BpmDj. It starts with explaining different techniques to measure tempo (energy enveloppe spectra, autocorrelation, peak detectors, ray-shooting and realtime tempo extraction). Beatgraphs are introduced, spectrum measurement based on the psychoacoustic bark scale is covered as well as echo characteristic, rythm and composition extraction. SVD / PCA analysis is used to map a multidimensional spectrum space to 3 primary colors
BpmDj Meta-Data Extraction
February 2005 – presented at the Computer Science Department, University of Tromso – Werner Van Belle

The presentation covers all meta data extraction aspects in BpmDj. It starts with explaining different techniques to measure tempo (energy enveloppe spectra, autocorrelation, peak detectors, ray-shooting and realtime tempo extraction). Beatgraphs are introduced, spectrum measurement based on the psychoacoustic bark scale is covered as well as echo characteristic, rythm and composition extraction. SVD / PCA analysis is used to map a multidimensional spectrum space to 3 primary colors
P53 Image Denoising
February 2005 – presented at Norut IT as part of a project delivarable in cooperation with the Hematyology Section, University Bergen – Werner Van Belle

Overview of Bioinformatics at Norut IT
December 2004 – presented at the Veterinaer Institute, Tromso – Werner Van Belle

Biological Systems, Modeling & Simulation
August 2004 – presented at PROBE, the proteomics unit in Bergen – Werner Van Belle

Denoising of MALDI-TOF Mass Spectra
August 2004 – presented at the Computational Biology Unit (CBU), Bergen – Werner Van Belle

Biological Systems, Modeling & Simulation
August 2004 – presented at Norut IT, Tromso – Werner Van Belle

BpmDj Meta-Data Extraction
May 2004 – presented at the Earth Observation Group, Norut IT, Tromso – Werner Van Belle

From Connection-less over Connection-oriented towards…
January 2004 – presented at the Computer Science Departement, University of Tromø – Werner Van Belle

The Development of an Intelligent Concurrency Adaptor in order To Mediate The Differences between Conflicting Concurrency Interfaces
November 2003 – presented at the Vrije Universiteit Brussel, Public Session PhD Defense – Werner Van Belle

The Development of an Intelligent Concurrency Adaptor in order To Mediate The Differences between Conflicting Concurrency Interfaces
September 2003 – presented at the Vrije Universiteit Brussel, Closed Session PhD Defense – Werner Van Belle

Adaptor Generation for Concurrency Interface Conflicts by Means of Petri-nets and Genetic Algorithms
November 2002 – presented at Institut für Informatik (IAM), Universität Bern, Bern – Werner Van Belle

Automatic Adaptor Generation by Means of Evolution
May 2002 – presented at the Programming Technology Lab, Vrije Universiteit Brussels (PROG/VUB) – Werner Van Belle

The Component Architecture
April 2002 – presented at Siemens ATEA, Herentals – Werner Van Belle, David Urting

Building Components for Mid-Scale Embedded Systems: What Kind of Infrastructure do we Need: the Component Architecture
December 2001 – presented at Philips TASS, Brussels – Werner Van Belle

Borg, a Way to hapiness in distributed environments: differences with pico
December 2001 – presented at the Prototype Meeting at the Programming Technology Lab (PROG), Vrije Universiteit Brussel – Werner Van Belle

A ‘consolidation’ effort to get other members of prog to develop using the same tools and code base. Needless to say, it didn’t work out.
The Component System a bit further
October 2001 – presented at the SEESCOA Meeting Brussels – Werner Van Belle

Concurrency Guarding and Error Resolution: Genetic Programming
September 2001 – presented at the Programming Technology Lab, Computer Science Department, Vrije Universiteit Brussel – Werner Van Belle

About the lack of concurrency control primitives in Open Distributed Object Systems
June 2001 – presented at the European Conference on Object Orientd Programming 2001 (ECOOP 2001), workshop 17 on object oriented distributed systems, Budapest – Werner Van Belle

This presentation discusses the problems one faces in open distributed with respect to modular and composable concurrency guarding primitives.
The Pitfalls of the Connectionless Component Model
April 2001 – presented at Inno.com, Beersel – Werner Van Belle

Experiences in Mobile Computing: The CBorg Mobile Multi-Agent System
March 2001 – presented at the Conference on Technology of Object Oriented Languages and Systems, Eastern Europe (TOOLSEE 2001): Components for Mobile Computing, Zurich – Werner Van Belle, Karsten Verelst, Johan Fabry, Theo D’Hondt

This presentation reports on our experiences in the field of mobile components. In the past 4 years we developed a mobile component system, which allowed us to experiment with code mobility in distributed systems. These experiments have given us a unique opportunity to study two major issues in mobile component systems. The first issue is how to develop and provide a robust mobile component architecture. The second issue is how to write code in these kinds of systems. This paper discusses our experience in both of the above.
The Component System so far
January 2001 – presented at the SEESCOA User Group Meeting in Gent – Werner Van Belle, David Urting

Glueing Components in Wide Area Neworks
November 2000 – presented at the Programming Technology Lab, Computer Science Department, Vrije Universiteit Brussel – Werner Van Belle

A very cool concept also known as the return continuation
The CBorg Mobile Multi Agent System: An Overview
August 2000 – presented at the Company Contact Presentation organized by the Programming Technology Lab (PROG), Vrije Universiteit Brussel – Werner Van Belle, Karsten Verelst, Theo D’Hondt

Agent Mobility and Reification of Computational State, an Experiment in Migration
June 2000 – presented at the European Conference on Object Oriented Programming 2000 (ECOOP 2000), Cannes – Werner Van Belle, Theo D’Hondt

The CBorg Mobile Multi Agent System: An Overview
June 2000 – presented at Ecole des Mines de Nantes – Werner Van Belle, Karsten Verelst, Johan Fabry, Theo D’Hondt

A mobile multi-agent is an active autonomous software component that is able to communicate with other agents; the term mobile refers to the fact that an agent can migrate to other agent systems, thereby carrying its program code and data along. Regarding the expression mobile multi-agent system there is a little confusion. A multi-agent system in AI denotes a software system that simulates the behavior of large groups of interacting agents, without focusing on the distribution aspect of these systems. In the world of distributed computing, a mobile multi-agent system is a distributed environment which multi-agents can be written in. Mobile multi-agent systems can be implemented as a library in an existing language, or as a special state of the art language that has high level primitives to write agents in. Nevertheless, whether one conceives an agent system as a library or as a language, the main goal is the same. A mobile multi-agent system should offer as much flexibility as possible to the agent programmer while hiding the underlying network architecture as much as possible. At PROG we developed a system called CBorg, which is a state of the art mobile multi agent system. The system is build upon Pico , written by Theo D’hondt. The talk will consist of two parts. The first part discusses the use of the system, illustrating the powerful semantics of the Pico language and detailing the current extensions made to create CBorg. These include a location transparent routing mechanism, a number of synchronization primitives and finally strong migration. The second part of the talk will focus on the Pico and CBorg virtual machine. We will explain the necessary requirements for the virtual machine to allow strong migration, give an overview of the routing mechanism and explain what kind of synchronization and concurrency handling mechanism is used. Finally we will discuss our current work on continuations which should nicely glue all these concepts together.
Glue Components & Meta-level Architectures
May 2000 – presented at the programming technology lab, computer science department, Vrije Universiteit Brussel – Philippe Demaecker

A Thesis Presentation given at PROG. Very interesting research
Software Modelling Techniques for Embedded Systems
April 2000 – presented at a SEESCOA User Meeting – Werner Van Belle, Tom Toutenel, David Urting, Theo D’Hondt, Yolande Berbers, Viviane Jonckers, Stefan Van Baelen, Koen Debosschere, Chris Luyten

Components
March 2000 – presented at a SEESCOA User Meeting – Werner Van Belle, Tom Toutenel

Location Transparent Routing in Mobile Multi Agent Systems: Merging Name Lookups with Routing
December 1999 – presented at Conference on Future Trends in Distributed Computing (FTDCS 1999), Cape Town – Werner Van Belle, Karsten Verelst, Theo D’Hondt

Applicability Of UML/RT to Embedded Systems
November 1999 – presented at a SEESCOA Meeting at the Katholic University of Leuven – Werner Van Belle, Tom Toutenel

Control Flow Structures in Distributed Programs
November 1999 – presented at the programming technology lab, computer science department, Vrije Universiteit Brussel – Philippe Demaecker, Werner Van Belle

Gevalstudie Mobiele Multi-Agent Systemen: CBorg
October 1999 – presented at the SEESCOA Kick-off meeting, Computer Science Department (DISTRINET), Katholic University Leuven (KUL) – Werner Van Belle

Very Short Introduction to PROG
October 1999 – presented at the SEESCOA Kickoff meeting at the Catholic University Of Leuven – Werner Van Belle

A presentation given in Leuven at the SEESCOA Kick-off meeting. It is a very short introduction to Prog and the Department Computer Science at the Universitry Brussels. Remarkable here was that I seemed to spend more time on the background than the actual content itself 🙂 So just out of nostalgic purposes it is here. It was also not such a very good introduction to what the lab actually does, but anyway, we only had 4 minutes to make the presentation and 5 to present it. So, bottomline is: one background good spend 🙂
Gevalstudie Mobiele Multi-Agent Systemen: CBorg
September 1999 – presented at the FKFO meeting Antwerpen, Universitaire Installing Antwerpen – Werner Van Belle, Theo D’Hondt

Gevalstudie Mobiele Multi-Agent Systemen: CBorg
December 1998 – presented at the Katholieke Universiteit Leuven – Werner Van Belle, Theo D’Hondt

Gevalstudie Mobiele Multi-Agent Systemen: CBorg
December 1998 – presented at the FKFO meeting Antwerpen, Universitaire Instelling Antwerpen – Werner Van Belle

Positie Optimalisaties Mobiele Multi-Agent Systems: stand van zaken
November 1998 – presented at the Programming Technology Lab, Computer Science Department, Vrije Universiteit Brussel – Werner Van Belle