Research Articles

On distances between energy vectors
January 2017 – Werner Van Belle


An audio-vector describes for each frequency band how much energy we find at a specific position. In this paper we investigate the problem of comparing audio-vectors. We present a new method to compare real or complex energyvectors.
Zathras-1: time stretching audio through matrix-multiplication
November 2015 – Werner Van Belle


Zathras-1 extends BpmDj with a time stretcher that does not modify the pitch of the sound. This article describes how we implemented a sinusoidal modelling by measuring the amplitude and phase envelope of peaks.
FlowDb – getting rid of SQL
August 2014 – Werner Van Belle


This article focuses on the creation of FlowDb, an object oriented database that replaced SQL as backend in BpmDj. The database supports a transactional memory, virtual paging, garbage collection, garbage compaction, change listeners and petri-nets.
Training a Distance Metric
June 2014 – Werner Van Belle


This article explains how to train a distance metric in a multi-dimensional space.
Two Solutions to Harmonics due to short Envelope Fades.
January 2014 – Werner Van Belle


When dealing with audiosignals, often a tone of known frequency must be faded in or out (for e:g suppression of ringing, or attack times of compressors). When the period used to fade in the tone is sufficiently short, overtones are created. We prove that only two fade in times canbe used to only have harmonics that are octaves apart. The first solution (a fade time of half the primary frequency) produces an harmonic 1 octave higher. The second (a fade time of a sixth period), produces harmonics at 1 octave and 2 octaves higher. No other solutions exist.
Estimating the phase of a subbin sine from a local group of frequencies.
January 2014 – Werner Van Belle


When sinusoidal modelling a peak from a fourier spectrum, the phase must often be estimated. Under specific FFT windows these phases can be interpolated. However, in the general sense such interpolation techniques do not work very well. We present a general purpose solution to estimate the frequency of a wave by looking at its neighboring bins.
Beatgraphs: Music at a Glance
February 2013 – Werner Van Belle


Explained are beatgraphs, a music visualisation technique that allows one to see rhythmical patterns, recognize frequencies and observer the musical structure in the blink of an eye.
Detecting Transients in Audio
August 2012 – Werner Van Belle


Explained is a transientdetector to detect changes in frequency or volume.
Fast Exponential Envelopes
April 2012 – Werner Van Belle


Computing ADSR curves is a task any synth has to deal with. In this article I present a method to compute exponential patches quickly and incrementally. Namely by multiplying the last envelopevalue with a constant multiplier (r) and then adding a constant delta (d) to it. This solution is used in the Audiotool Heisenberg synth to generate ADSR curves for pitch as well as amplitude.
Enhancing Loudness and Avoiding Compressor Carnage
March 2012 – Werner Van Belle


Explains a- the cause of compressor carnage, which is the endless stream of ticks produced by many commercial compressors. b- how a hold circuit combined with a Bessel filter can mediate this problem. c- how to efficiently implement the required attack and hold lines using the Titchmarsh convolution theorem. d- how a loudness enhancer with very short attack/decay times can be created using the above methods.
Correlation analysis of p53 protein isoforms with NPM1/FLT3 mutations and therapy response in acute myeloid leukemia
March 2012 – Nina Ånensen, Werner Van Belle, S.M. Hjelle, Ingvild Haaland, E Silden, Jean-Christophe Bourdon, R Hovland, K Taskén, S Knappskog, P.E. Lønning, Oystein Bruserud, Bjørn Tore Gjertsen


The wild-type tumor-suppressor gene TP53 encodes several isoforms of the p53 protein. However, while the role of p53 in controlling normal cell cycle progression and tumor suppression is well established, the clinical significance of p53 isoform expression is unknown. A novel bioinformatic analysis of p53 isoform expression in 68 patients with acute myeloid leukemia revealed distinct p53 protein biosignatures correlating with clinical outcome. Furthermore, we show that mutated FLT3, a prognostic marker for short survival in AML, is associated with expression of full-length p53. In contrast, mutated NPM1, a prognostic marker for long-term survival, correlated with p53 isoforms ? and ? expression. In conclusion, p53 biosignatures contain useful information for cancer evaluation and prognostication.
Fast Tempo Measurement: a 6000x sped-up autodifferencer.
July 2011 – Werner Van Belle


The autodifference is often used to measure the tempo of music. This technique requires quadratic time, with one parameter being the tempo range under investigation and the second being the songlength. In this paper we describe a number of improvements to this analysis technique to reduce the number of required calculations with a factor of 6272. We also report on a solution for the problem of autodifferencebiases and finally we explain a new solution to resolve tempo harmonic for 4/4 rhythms.
Bitslicing: Nearest Neighbors with <30% Data Access for 1000+ Dimensional Uniform Spaces
June 2011 – Werner Van Belle


We describe a space partitioning scheme to store and retrieve high dimensional integer data (>1000 dimensions). The search structure is similar to hypertrees but retrieves the k nearest neighbors by accessing only ~30% of the search space. To make this possible, a breadth-first search culls away large areas of the space by using an incremental calculation of the minimal distance between each subspace and the point-of-interest.
Classification of Rhythmical Patterns
September 2010 – Werner Van Belle


We classify rhythmical patterns using a variety of classifiers and report on the results. In summary: they do not directly seem useful to help us classify tempo multiples properly. However, as a side effect of this research we found a hint that a flux based measurement (how fast energy changes in a song) might be a useful idea to classify music into slow or fast classes.
Rhythm Pattern Extraction
August 2010 – Werner Van Belle


Once the tempo of music is known, that information can be used to extract rhythmical information. This is done by extracting all consecutive measures and combining them into an average measure. During this process a number of normalisation factors should be performed. In this paper we discuss the removal of sound color, echoes and rotational offsets.
Biasremoval from BPM Counters and an argument for the use of Measures per Minute instead of Beats per Minute
July 2010 – 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 the bias introduced through coherent noise. This bias is unique for each song and without its removal reported tempos might be off by a multiplication factor. This article 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. 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.
A Combined Ex Vivo and In Vivo RNAi Screen for Notch Regulators in Drosophila Reveals an Extensive Notch Interaction Network
May 2010 – Abil Saj, Zeynep Arziman, Denise Stempfle, Werner Van Belle, Ursula Sauder, Thomas Horn, Markus Dürrenberger, Renato Paro, Michael Boutros, Gunter Merde


Notch signaling plays a fundamental role in cellular differentiation and has been linked to human diseases, including cancer. We report the use of comprehensive RNAi analyses to dissect Notch regulation and its connections to cellular pathways. A cell-based RNAi screen identified 900 candidate Notch regulators on a genome-wide scale. The subsequent use of a library of transgenic Drosophila expressing RNAi constructs enabled large-scale in vivo validation and confirmed 333 of 501 tested genes as Notch regulators. Mapping the phenotypic attributes of our data on an interaction network identified another 68 relevant genes and revealed several modules of unexpected Notch regulatory activity. In particular, we note an intriguing relationship to pyruvate metabolism, which may be relevant to cancer. Our study reveals a hitherto unappreciated diversity of tissue-specific modulators impinging on Notch and opens new avenues for studying Notch regulation and function in development and disease.
Active Objects: an Efficient Space/Time Separation of Concurrent Tasks through Active Message Queues
February 2010 – Werner Van Belle


This article focuses on Active Message Queues, an abstraction which I created between 1998 and 2006. The abstraction seperates various tasks through means of a message queue whereby the queue will activate the underlying task whenever a new message comes in. Using this abstraction it becomes possible to have time speration between threads (they will not influence each others execution/not wait for each other), space separation (different tasks will not write into each others data space), scheduling independence (we can run multiple tasks with one, two or more processes), syntactical separation (each task is declared in its own source file) and finally lockfree programming, which thus also leads to a noticable lack of deadlock and race-conditions
FlowPipes: Isolating and Rejoining Data Tokens in the context of Data Analysis Pipelines
December 2009 – Werner Van Belle


FlowPipes is a data processing abstraction that helps to integrate data analysis modules into a data analysis pipeline. The infrastructure provides automatic joining of relevant data as well as rejoining of the analysis output into the overall flow. FlowPipes explicitely separates a meta-level (in which tokens drive the computation) from an object-level (where commands process raw data). The computation is driven through tokenmatching: when data-tokens can be combined then the computation continues along that path. The resulting pipelines are data-driven and programatorically functionially oriented. We demonstrate our work on two bioinformatics examples: a real time PCR calculator and an image correlation tool.
Anchoring gaze when categorizing faces sex: Evidence from eye-tracking data
August 2009 – Line Sæther, Werner Van Belle, Bruno Laeng, Tim Brennen, Øvervoll Morten


Previous research has shown that during recognition of frontal views of faces, the preferred landing positions of eye fixations are either on the nose or the eye region. Can these findings generalize to other facial views and a simpler perceptual task? An eye-tracking experiment investigated categorization of the sex of faces seen in four views. The results revealed a strategy, preferred in all views, which consisted of focusing gaze within an infraorbital region of the face. This region was fixated more in the first than in subsequent fixations. Males anchored gaze lower and more centrally than females.
Methods for analyzing deep sequencing expression data: constructing the human and mouse promoterome with deepCAGE data.
July 2009 – Piotr J. Balwierz, Piero Carninci, Carsten Daub, Jun Kawai, Werner Van Belle, Christian Beisel, Erik Van Nimwegen


With the advent of ultra high-throughput sequencing technologies, increasingly researchers are turning to deep sequencing for gene expression studies. Here we present a set of rigorous methods for normalization, quantification of noise, and co-expression analysis of deep sequencing data. Using these methods on 122 cap analysis of gene expression (CAGE) samples of transcription start sites, we construct genome-wide ‘promoteromes’ in human and mouse consisting of a three-tiered hierarchy of transcription start sites, transcription start clusters, and transcription start regions.
Pattern Analysis of Human Histone Methylations and Acetylations through Cross Correlation Maps
February 2009 – Werner Van Belle


This paper investigates how to properly calculate relations between multiple genomic tracks produced by deep sequencing technology. Based on a cross-correlation analysis we measure relations between peak and peaks, peaks and areas and areas and areas over short or long distances. Using this method we revisit previously published acetylation and methylation patterns and present 41 maps, one for each of the measured histone modifications.
Propagating Differential Gene Expression in Protein Interaction Networks
February 2008 – Werner Van Belle, Nancy Gerits, Ugo Moens


In a differential gene experiment, a cell perturbation can be measured on a microarray before and after the perturbation. The information from these microarrays can then be used to inference genetic pathways and protein-protein interaction networks. In this paper we reverse this idea somewhat and measure a cell perturbation through microarrays and then rely on a protein interaction map to assess which proteins are most likely influenced by the specific perturbation. This in turn helps to elucidate the functional effect the perturbation has on the cell system. The first part of the paper focuses on the propagation model we developed to obtain this information. The second part of the paper reports on a specific experiment that was driven by the interpretation we obtained through such a gene influence network. We applied a PC12 cell line that allows doxocyclin-dependent expression of constitutive active mitogen-activated protein kinase-activated protein kinase (MAPKAPK5 or MK5) to compare the transcription pattern in the presence or absence of doxocyclin by microarray analysis. The subsequent protein network analysis suggested that MK5 would affect anxiety. This finding was then confirmed by means of transgenic mice on an elevated plus maze and in a light-dark box.
Transgenic mice expressing constitutive active MAPKAPK5 display gender-dependent differences in exploration and activity
November 2007 – Nancy Gerits, Werner Van Belle, Ugo Moens


Background: The mitogen-activated protein kinases, MAPKs for short, constitute cascades of signalling pathways involved in the regulation of several cellular processes that include cell proliferation, differentiation and motility. They also intervene in neurological processes like fear conditioning and memory. Since little remains known about the MAPK-Activated Protein Kinase, MAPKAPK5, we constructed the first MAPKAPK knockin mouse model, using a constitutive active variant of MAPKAPK5 and analysed the resulting mice for changes in anxiety-related behaviour. Methods: We performed primary SHIRPA observations during background breeding into the C57BL/6 background and assessed the behaviour of the background-bred animals on the elevated plus maze and in the light-dark test. Our results were analysed using Chi-square tests and homo- and heteroscedatic T-tests; Results: Female transgenic mice displayed increased amounts of head dips and open arm time on the maze, compared to littermate controls. In addition, they also explored further into the open arm on the elevated plus maze and were less active in the closed arm compared to littermate controls. Male transgenic mice displayed no differences in anxiety, but their locomotor activity increased compared to non-transgenic littermates; Conclusions: Our results revealed anxiety-related traits and locomotor differences between transgenic mice expressing constitutive active MAPKAPK5 and control littermates.
An Adaptive Filter for the Correct Localization of Subimages: FFT based Subimage Localization Requires Image Normalization to work properly
October 2007 – Werner Van Belle


How to find subimages with the fast fourier transform ? It is well documented how Fourier transforms can speed up image alignment. The cross-correlation of an image can be calculated as the inverse Fourier transform of the inproduct of the Fourier transform of the first image and the conjugated Fourier transform of the second image. However, when one tries to extend this useful relation to locate subimages in larger images we find that it no longer performs properly. In this document we describe a fast normalization method that makes it possible to use the standard Fourier based cross-correlation technique for subimage localization. The technique relies on a low pass filtering followed by an adaptive contrast enhancement.
Confidence Intervals on Microarray Measurement of Differentially Expressed Genes: A Case Study on the effects of MK5, TAF4 and FKRP on the transcriptome
May 2007 – Werner Van Belle, Nancy Gerits, Kirsti Jakobsen, Vigdis Brox, Marijke Van Ghelue, Ugo Moens


To perform a quantitative analysis with gene-arrays, one must take into account inaccuracies (experimental variations, biological variations and other measurement errors) which are seldomly known. In this paper we investigated amplification and noise propagation related errors by measuring intensity dependent variations. Based on a set of control samples, we create confidence intervals on up and down regulations. We validated our method through a qPCR experiment and compared it to standard analysis methods (including loess normalization and filtering methods based on genetic variability). The results reveal that experimental variability and amplification related errors are a major concern that should be accounted for.
Correlation between the inproduct and the sum of absolute differences is -0.8485 for uniform sampled signals on [-1:1]
November 2006 – Werner Van Belle


While investigating relations between various comparators and the inproduct we found that the inproduct correlates strongly towards the absolute difference when the domain from which the values are taken come from a uniform distribution on [-1:1]. This useful relation might help to speed up block comparison in content databases and can be used as a valuable tool to estimate the inproduct based on the absolute difference and vice versa
Antenna Finding and Interpolation/Extrapolation of Signal Strength
November 2006 – Werner Van Belle


This article investigates the possibility to determine the position of a WiFi antenna by sampling the environment using a standard wireless card. The method is based on a search algorithm (simulated annealing) in a renormalized antenna model. The article further documents a sample acquisition phase in which a static wireless card measured the local environment. This might allow the integration of various oscillatory phenomena into the model.
Ambient Actors as a Formalism for Ubiquitous and Mobile Computing
October 2006 – Werner Van Belle, Jessie Dedecker


Networks formed by the interconnection of mobile wireless devices are often called mobile ad-hoc networks. Such networks are highly volatile because communication partners can move in and out of range. This leads to unwanted interruptions of communication sessions, which in turn complicates the development of software that relies on a long term distributed state. Standard solutions seem inadequate because all too often they assume an inherent client server role or they assume a bound latency on communication sessions. One cannot make such assumptions in mobile ad-hoc networks. In order to resolve this matter a programming model must support the ability to pick up previous communication sessions; it must remember the devices/resources it has seen in the past; and it must provide a realistic approach towards concurrency, taking into account the limitations of a peer to peer ad-hoc network. This article presents the ambient actor model which extends the operational semantics of the actor model to capture the limitations posed by this novel paradigm
Dj-ing under Linux with Bpmdj
October 2006 – Werner Van Belle


BpmDj provides the digital DJ with QT based tools to analyze, manage and play mp3/ogg songs. The program annotates all songs with psychoacoustic properties, including tempo, sound color, echo/delay information, rhythm information and composition information. The program relies on mplayer, ogg123 and/or mpg123 to decode songs while output is send through alsa or oss drivers. This article demonstrates beatmixing (how a DJ can superimpose two songs during longer times) and automixing (how BpmDj can mix music for you). It further sheds some light on the psychoacoustic properties used and a feature called beatgraphs.
Adaptive contrast enhancement of two-dimensional electrophoretic protein gel images facilitates visualization, orientation and alignment
October 2006 – Werner Van Belle, Gry Sjøholt, Nina Ånensen, Kjell-Arild Høgda, Bjørn Tore Gjertsen


Two-dimensional polyacrylamide gel electrophoresis (2-DE) is a powerful technique to discriminate post-translationally modified protein isoforms. However, all steps of 2-DE preparation and gel-staining may introduce unwanted artefacts, including inconsistent variation of background intensity over the entire 2-DE gel image. Background intensity variations limit the accuracy of gel orientation, overlay alignment and spot detection methods. We present a compact and efficient denoising algorithm that adaptively enhances the image contrast and then, through thresholding and median filtering, removes the gray-scale range covering the background. Applicability of the algorithm is demonstrated on immuno-blots, isotope labeled gels, and protein stained gels. Validation is performed in contexts of i) automatic gel orientation based on Hough transformation, ii) overlay alignment based on cross correlation and iii) spot detection. In gel-stains with low background variability, e.g. Sypro Ruby, denoising will lower the spot detection sensitivity. In gel regions with high background levels denoising enhances spot detection. We propose that the denoising algorithm prepares images with high background for further automatic analysis, without requiring manual input on a gel-to-gel basis.
Proteomics of p53 in Diagnostics and Therapy of Acute Myeloid Leukemia
July 2006 – Nina Ånensen, Ingvild Haaland, Clive D’Santos, Werner Van Belle, Bjørn Tore Gjertsen


The anti-oncogene TP53 is frequently mutated in human cancer, but in hematological malignancies this is a rare feature. In acute myeloid leukemia (AML) more than 90% of the patients comprise wild type TP53 in their cancer cells, but if TP53 is mutated or deleted the disease is often found to be chemoresistant. In this review we define proteomics of the oncogene product p53 as the study of proteins in the p53 regulating signaling networks, as well as the protein study of members of the p53 family itself. Various messenger RNA splice forms as well as a multitude of posttranslational modifications give a high number of protein isoforms in the p53 family. Some of the proteomic techniques allow detection of various isoforms, such as two-dimensional gel electrophoresis in combination with tandem mass spectrometry (MS/MS) and this methodology may therefore increasingly be used as a diagnostic tool in human disease. We introduce the p53 protein as an illustration of the complexity of post-translational modifications that may affect one highly connected protein and discuss the possible impact in AML diagnostics if the p53 profile is reflecting cell stress and status of signal transduction systems of the malignancy
Correlation analysis of two-dimensional gel electrophoretic protein patterns and biological variables
April 2006 – Werner Van Belle, Nina Ånensen, Ingvild Haaland, Oystein Bruserud, Kjell-Arild Høgda, Bjørn Tore Gjertsen


Background) Two-dimensional gel electrophoresis (2DE) is a powerful technique to examine post-translational modifications of complexly modulated proteins. Currently, spot detection is a necessary step to assess relations between spots and biological variables. This often proves time consuming and difficult when working with non-perfect gels. We developed an analysis technique to measure correlation between 2DE images and biological variables on a pixel by pixel basis. After image alignment and normalization, the biological parameters and pixel values are replaced by their specific rank. These rank adjusted images and parameters are then put into a standard linear Pearson correlation and further tested for significance and variance. Results) We validated this technique on a set of simulated 2DE images, which revealed also correct working under the presence of normalization factors. This was followed by an analysis of p53 2DE immunoblots from cancer cells, known to have unique signaling networks. Since p53 is altered through these signaling networks, we expected to find correlations between the cancer type (acute lymphoblastic leukemia and acute myeloid leukemia) and the p53 profiles. A second correlation analysis revealed a more complex relation between the differentiation stage in acute myeloid leukemia and p53 protein isoforms. Conclusion) The presented analysis method measures relations between 2DE images and external variables without requiring spot detection, thereby enabling the exploration of biosignatures of complex signaling networks in biological systems.
Method and Apparatus for Correlation Analysis of Electrophoretic Protein Patterns
April 2006 – Werner Van Belle, Bjørn Tore Gjertsen, Rebecca Katherine


A technique I used to analyze 2DE gels is currently being patented by TTO Nord AS (Technology Transfer Office). The patent application is filed in Great Brittain with application number GB0604906.8. This patent should never have been filed because: a) it is a mathematical technique b) there is no ‘innovative step’ involved (correlation is not new) c) one of the claims is completely for free without any evidence supporting a predicitve possibibilty and d) the patent content has been presented at many public fora before it was filed. The current description goes as follows: (EN) A method of obtaining clinical information about a cancer patient by analysing p53 in a sample taken from said patient. It comprises the steps of: a) providing an electrophoresis gelimage containing spatially resolved data indicative of the concentration of one or more p53 isoforms in said sample; b) providing a set of previously acquired correlation data that represents the correlation between a plurality of pixels in the type of image in step a) and the clinical information of interest; and b) using the correlation data to obtain said clinical information from the image. The invention also enables such correlation data to be used to predict a value for a clinical parameter from a gel image. (FR): L’invention concerne un procédé d’obtention d’informations cliniques concernant un patient cancéreux par analyse du gène p53 dans un échantillon prélevé sur ledit patient. Le procédé comprend les étapes consistant à : a) mettre à disposition une image d’un gel d’électrophorèse contenant des données à résolution spatiale représentant la concentration d’une ou de plusieurs isoformes du gène p53 dans ledit échantillon; b) mettre à disposition un groupe de données de corrélation préalablement acquises représentant la corrélation entre une pluralité de pixels dans le type d’image créée à l’étape a) et les informations cliniques d’intérêt; et c) utiliser les données de corrélation pour obtenir lesdites informations cliniques à partir de l’image. L’invention permet également d’utiliser ces données de corrélation pour prévoir la valeur d’un paramètre clinique à partir d’une image du gel.
Observations on spectrum and spectrum histograms in BpmDj
September 2005 – Werner Van Belle


BpmDj Is a program for DJ’s. It helps to select songs and play them. To achieve this the program relies on a number of signal processing techniques. One of the available techniques compares song spectra. Both a standard spectrum analysis is performed as well as a distribution analysis (‘echo’ characteristics). In this short article we describe how this property is calculated and how it is further used in song comparison. We also present a short analysis of the structure of the space using correlation techniques.
Artefacts in the Mass Spectra Output from MALDI-TOF and MALDI-TOF/TOF Machines
April 2005 – Werner Van Belle, Olav Mjaavatten


MALDI-TOF mass spectrometry is a well known and widely used technique to fingerprint and sequence proteins. A carefull investigation of the mass spectra output from unnamed machines shows a number of artefacts produced by the machines themselves. Because these artefacts complicate a number of procedures we present a number of preliminary techniques we developed to get rid of most of the artefacts.
Actors for Mobile Ad-Hoc Networks (The Ambient Actor Model)
August 2004 – Jessie Dedecker, Werner Van Belle


Wireless communication has a limited communication range which introduces two major problems, currently not captured in distributed middleware. First, they are less reliable than their wired variants and secondly they are extremely dynamic. Both problems complicate the development of mobile software. In this paper we extend the operational semantics of the actor model to capture these two properties. We do this by adding a single new concept to the model: the mailbox. This paper provides a foundation for new implementations of the actor language and frameworks that are usable in the context of wireless network environments.
The Failure of the Connectionless Model and the Vice Software Model
February 2004 – Werner Van Belle


We investigate a number of modularity problems that arise from using a connection-less software model. A connection-less approach leads naturally to an unmanaged use of direct references. This results in crucial software composition complications such as the inability to replace objects, difficulties to track outgoing messages, difficulties in writing glue-codehidden communication. As a solution to these problems, we present our connection oriented software model, Vice. Its reflective design makes that it surpasses existing architectures in its modularity, manageability and performance.
Actors for Pervasive Computing
June 2003 – Jessie Dedecker, Werner Van Belle, Wolfgang De Meuter


The position we defend in this paper is that asynchronous communication patterns are to be considered to be the main communication paradigm in pervasive applications and that the actor model, which relies on such a form of communication, needs extenstion to be practical useful.
Using Genetic Programming to Generate Protocol Adaptors for Interprocess Communication
March 2003 – Werner Van Belle, Tom Mens, Theo D’Hondt


As mobile devices become more powerful, interprocess communication becomes increasingly more important. Unfortunately, this larger freedom of mobility gives rise to unknown environments. In these environments, processes that want to communicate with each other will be unable to do so because of protocol conflicts. Although conflicting protocols can be remedied by using adaptors, the number of possible combinations of different protocols increases dramatically. Therefore we propose a technique to generate protocol adaptors automatically. This is realised by means of genetically engineered classifier systems that use Petri nets as a specification for the underlying protocols. This paper reports on an experiment that validates this approach.
Automatic Adaptor Generation by means of Genetic Algorithms
August 2002 – Werner Van Belle, Tom Mens, Theo D’Hondt


Mobile multi-agent systems can be seen as a basis for global peer-to-peer computing. This new computation paradigm becomes increasingly more important as mobile devices become more powerful. Unfortunately, an open internet environment is an excellent area for interface conflicts between agents that want to communicate with each other. Although conflicting interfaces can be remedied by using adaptors, the number of possible combinations of different interfaces increases dramatically. Therefore we propose a technique to generate interface adaptors automatically. This is achieved by means of genetically engineered classifier systems (a well-known genetic algorithm technique) that use Petri nets as a specification for the underlying interfaces. This paper reports on an experiment that validates this approach. For designers of mobile applications, our approach is an important step forward, since the task of the developer is shifted from the writing of adaptors to the specification of test scenarios.
An experiment in automatic Adaptor Generation by means of Genetic Algorithms / Ma2002
May 2002 – Werner Van Belle, Theo D’Hondt


Mobile multi agent systems can be seen as a basis for global peer to peer computing. Nevertheless such an open environment makes it difficult to write agents which can interface with other agents because it is an excellent area for interface conflicts. Conflicting interfaces can be remedied by using adaptors. It is possible to automatically generate those glue-adaptors by means of genetically engineered classifier systems, which use Petri-nets as a model for the underlying interfaces. This paper reports on an experiment that illustrates this approach.
An experiment in automatic Adaptor Generation by means of Genetic Algorithms / Aamas2002
April 2002 – Werner Van Belle, Theo D’Hondt


Mobile multi agent systems can be seen as a basis for global peer to peer computing. Nevertheless such an open environment makes it difficult to write agents which can interface with other agents because it is an excellent area for interface conflicts. Conflicting interfaces can be remedied by using adaptors. It is possible to automatically generate those glue-adaptors by means of genetically engineered classifier systems, which use Petri-nets as a model for the underlying interfaces. This paper reports on an experiment that illustrates this approach.
The Reflective Virtual Machine
June 2001 – Karsten Verelst, Werner Van Belle, Theo D’Hondt


We claim that current day reflective architectures do not offer sufficient functionality, and that new developments in computer science push us towards a stronger reflective model: reflective virtual machines. We have witnessed these shortcomings in the application domain of mobility. Strong mobility is very difficult to implement in today’s programming languages, mainly because of the inability to capture the program’s computational state. Therefore we propose a new reflective architecture, the reflective virtual machine, that offers sufficient support for applications in mobility. In this paper we will first describe the basic functionality a mobile agentplatform should offer. This shall be done using a solution to the malicious host problem as a case. After identifying these needs we will introduce an interpreter, the Reflective Virtual Machine, that offers sufficient reflection, so that mobile applications can be straightforwardly implemented.
Is Message Sending Good enough for Distributed Systems ? Communication and synchronisation revisited
June 2001 – Werner Van Belle, Karsten Verelst, Kristof Van Buggenhout, Theo D’Hondt


This position paper looks at current day distributed object systems. In most of these systems we see a basic building primitive which offers us the needed distribution capabilities. Most of the time this is a synchronous or asynchronous message send. We argue that this is not good enough for distributed systems. Choosing the ‘message-send’ as primitive operation creates too many problems and creates the illusion that writing distributed programs is as easy as ‘calling an object’. This is not true. In this paper we would like to present our ideas about this and hope to provoke some discussion about them.
Experiences in Mobile Computing: The CBorg Mobile Multi-Agent System
March 2001 – Werner Van Belle, Johan Fabry, Karsten Verelst, Theo D’Hondt


This paper 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.
BPM Measurement of Digital Audio by Means of Beat Graphs & Ray Shooting
December 2000 – Werner Van Belle


In this paper we present a) a novel audio visualization technique, called beat-graphs and b) a fully automatic algorithm to measure the mean tempo of a song with a very high accuracy. The algorithm itself is an easy implementable offline search algorithm that looks for the tempo that best describes the song. For every investigated tempo, it steps through the song and compares the similarity of consecutive pieces of information (bass drum, a hi-hat, …). Its accuracy is two times higher than other fully automatic techniques, including Fourier analysis, envelope-spectrum analysis and autocorrelation.
The CBorg Mobile Multi-Agent System
October 2000 – Werner Van Belle, Karsten Verelst, Theo D’Hondt


This paper describes the design behind an experimental mobile multi agent system we have built over the past 4 years. The architecture itself is called Cborg. The objective was to create an architecture which can be easily used to implement intelligent autonomous agents in a wide area network.
Towards a unified terminology for component-based development
June 2000 – Stefan Van Baelen, David Urting, Werner Van Belle, Viviane Jonckers, Tom Holvoet, Yolande Berbers, Karel De Vlaminck


Component-oriented programming and component-based development have become rather mature software development approaches in the last years, with both quite good conceptual and technological support. In spite of this rapid growth, the concepts used when talking about components are not always well-defined and can lead to misconception, misunderstanding and confusion. Although the general, abstract definition of a component, as defined at WCOP96: A software component is a unit of composition with contractually specified interfaces and explicit context dependencies only. A software component can be deployed independently and is subject to composition by third parties. is widely accepted, a lot of confusion can arise when characteristics of components are described.
Agent mobility and Reification of Computational State
June 2000 – Werner Van Belle, Theo D’Hondt


This paper describes an experiment with mobility in multi-agent systems. The setting is a virtual machine that supports reification of the computational state of a running process. The objective is to investigate how this feature facilitates telescripting and to speculate on how languages like Java should evolve to include the resulting notion of strong migration.
Communication & Synchronization in Mobile Multi-Agent Systems, CSP Revisited
March 2000 – Karsten Verelst, Werner Van Belle, Theo D’Hondt


We have created a communication model by extending an imperative language with 4 simple communication primitives. It turns out that these four primitives are very versatile and cover a large portion of existing communication protocols such as CSP and Actor systems. Our communication model offers both synchronous and asynchronous communication in an easy-to-use communication model. This is achieved by extending an imperative language with 4 communication primitives. Because we desire asynchronous means of communication, our main communication primitives, send and receive, both work asynchronous. We have also included a sync-primitive that synchronizes two processes, and thus synchronous communication can be simulated. The fourth primitive, newagent, is used to create new agents. In this document we first introduce these communication primitives with the help of some examples. Next we give an example that demonstrates the use of our communication model. This example already gives a hint of the great expressive power of our communication language and in the last part of this document we show that our communication model can simulate the behaviour of more well-known protocols, such as CSP, Pi-calculus, actor-systems, RMI and others.
Location Transparent Routing in Mobile Agent Systems – Merging Name Lookups with Routing
December 1999 – Werner Van Belle, Karsten Verelst, Theo D’Hondt


Telecommunication systems these days are moving from static wide area component structures towards highly dynamic mobile infrastructures. This shift requires new algorithms to interconnect these mobile entities/components and route messages between them. In this paper we describe a naming and routing algorithm which can be used in fine-grained mobile component systems. As a case we use a homogenous environment of mobile multi-agent systems, which executes agents as they pass by.
Reinforcement Learning as a Routing Technique for Mobile Multi Agent Systems
January 1998 – Werner Van Belle


Nowadays, a number of mobile multi agent systems are designed and implemented at a variety of research labs. Most of these systems suffer from a major problem: location transparency. Most systems do not implement this, and the few of them which do are not robust towards changes in the network topology and/or absolutely imperformant in wide area networks. In this paper we present a naming scheme and a location transparent routing algorithm, based upon reinforced Q learning, which is robust and performant.
Prototype Based agents for the Web
September 1997 – Wolfgang De Meuter, Kris De Volder, Werner Van Belle, Tom Tourwe, Theo D’Hondt