Lars Kai Hansen

Theorems on Positive Data: On the Uniqueness of NMF (2008)

for difference between two groups of

Hans Laurberg, Mads Græsbøll Christensen, Mark D. Plumbley, Lars Kai Hansen, Søren Holdt Jensen

We investigate the conditions for which nonnegative matrix factorization (NMF) is unique and introduce several theorems which can determine whether the decomposition is in fact unique or not. The...

Theorems on Positive Data: On the Uniqueness of NMF (2008)

Hans Laurberg, Mads Græsbøll Christensen, Mark D. Plumbley, Lars Kai Hansen, Søren Holdt Jensen

We investigate the conditions for which nonnegative matrix factorization (NMF) is unique and introduce several theorems which can determine whether the decomposition is in fact unique or not. The...

Discovering Music Structure via Similarity Fusion (2007)

Arenas-Garcia, Jeronimo, Parrado-Hernandez, Emilio, Meng, Anders, Hansen, Lars Kai, Larsen, Jan

Automatic methods for music navigation and music recommendation exploit the structure in the music to carry out a meaningful exploration of the "song space''. To get a satisfactory performance from...

Modeling of BrainMap data (2007)

Finn Arup Nielsen, Lars Kai Hansen

We apply machine learning techniques in the form of Gaussian mixture models to functional brain activation data. The dataset was extracted through the WWW interface to the BrainMap TM

Universal Distribution of Saliencies for Pruning in Layered Neural Networks (2007)

J. Gorodkin, Lars Kai Hansen, L. K. Hansen, Benny Lautrup, S. A. Solla

A better understanding of pruning methods based on a ranking of weights according to their saliency in a trained network requires further information on the statistical properties of such saliencies....

Address for correspondence: (2007)

Lars Kai Hansen, Finn Arup Nielsen, Finn Arup Nielsen

We describe a content-based image retrieval technique for nding related functional neuroimaging experiments by voxelization of sets of stereotactic coordinates in Talairach space, comparing the...

Exploring fMRI data (2007)

Lars Kai Hansen, Finn Fkrup Nielsen, Jan Larsen, Mathematical Modelling

We use a Bayesian framework to detect periodic components in fMRI data. The resulting detector is sensitive to periodic components with a flexible number of harmonics and with arbitrary amplitude and...

volumes of interest via tbe (2007)

Finn Arup Nielsen, Lars Kai Hansen

Automatic anatomical labeling of Talairach coordinates and generation of

Brain Mapping by Positron Emission Tomography 1 1. Brain Mapping by Positron Emission Tomography (2007)

Peter Toft, Peter Alshede Philipsen, Lars Kai Hansen, Ulrik Kjems

In this paper the basis of PET (Positron Emission Tomography) is reviewed, and it is shown that the measured signals can be modelled as the Radon transform of the desired spatial distribution of,...

Running title: Feature-space clustering Address for correspondence: (2007)

Lars Kai Hansen, Matthew G. Liptrot, Egill Rostrup, Cyril Goutte, Cyril Goutte

Clustering fMRI time series has emerged in recent years as a possible alternative to parametric modelling approaches. Most of the work has been so far concerned with clustering raw time series. In...

DESIGN AND EVALUATION OF NEURAL CLASSIFIERS Mads Hintz-Madsen, Morten (2007)

Informatics and Mathematical Modeling (2007)

Abstract- In this paper we propose a method for design of feed-forward neural classifiers based on regularization and adaptive architectures. Using a penalized maximum likelihood scheme we derive a...

Jan Larsen, Claus Svarer, Lars Nonboe Andersen, Lars Kai Hansen

Rivals and Personnaz (Rivals & Personnaz, 2000) mainly concerns estimation of con dence intervals (or error bars) for neural network prediction models trained by least squares, but also the use...

2 (2007)

Abstract. In this paper we address the important problem of optimizing regularization parameters in neural network modeling. The suggested optimization scheme is an extended version of the recently...

DISCRIMINATION OF CYLINDERS WITH DIFFERENT WALL THICKNESSES USING NEURAL NETWORKS AND SIMULATED DOLPHIN SONAR SIGNALS (2007)

Lars Nonboe Andersen, Whitlow Au, Jan Larsen, Lars Kai Hansen

Abstract. This paper describes a method integrating neural networks into a system for recognizing underwater objects. The system is based on a combination of simulated dolphin sonar signals,...

and (2007)

Ole Winther, Lars Kai Hansen

We develop mean eld approaches for probabilistic independent component analysis (ICA). The sources are estimated from the mean of their posterior distribution and the mixing matrix (and noise level)...

1 1. Regularization of Neural Networks (2007)

Jan Larsen, Lars Kai Hansen, Claus Svarer

Neural networks are flexible tools for nonlinear function approximation and by expanding the network any relevant target function can be approximated [6]. The risk of overfitting

Time Series Processing (2007)

Torben L. Fog, Jan Larsen, Lars Kai Hansen

We study training and generalization for multi-variate time series processing. It is suggested to used a quasi maximum likelihood approach rather than the standard sum of squared errors, thus taking...

Are all e-customers alike? (2007)

Jens Hørlück, Jan Larsen, Lars Kai Hansen, Torben Christiansen

Contemporary marketing is based on an idea that life-style determines activity patterns. Based on these ideas, customers are segmented and marketing activity is based on these segments. In relation...

SONAR DISCRIMINATION OF CYLINDERS FROM DIFFERENT ANGLES USING NEURAL NETWORKS (2007)

Lars Nonboe Andersen, Whitlow Au, Jan Larsen, Lars Kai Hansen

This paper describes an underwater object discrimination system applied to recognize cylinders of various compositions from different angles. The system is based on a new combination of simulated...

y (2007)

Cyril Goutte, Lars Kai Hansen

Both authors wish to acknowledge stimulating discussions with Jan Larsen. Regularization with a pruning prior We investigate the use of a regularization prior and its pruning properties. We...

Active Surface Models for Brain Imaging (2007)

Mattias Ohlsson, Peter Toft, Lars Kai Hansen, Finn Arup Nielsen

This paper presents a generic approach for surface modelling of 3D objects from volume data. Our strategy is to shape a closed surface of connected triangles to match the edge of the object. The...

EARLY STOP CRITERION FROM THE BOOTSTRAP ENSEMBLE (2007)

Lars Kai Hansen, Jan Larsen, Torben Fog

This paper addresses the problem of generalization error estimation in neural networks. A new early stop criterion based on a Bootstrap estimate of the generlization error is suggested. The estimate...

Restoring functional PET Images using Anatomical MR Images (2007)

Peter Alshede Philipsen, Ulrik Kjems, Peter Toft, Lars Kai Hansen

In this paper we present a Bayesian method to enhance functional 3D PET images using apriori knowledge about the brain anatomy obtained from 3D MR images. We use a Markov Random Field as a prior...

Neural Networks (Computer). Address for correspondence: (2007)

Lars Kai Hansen, Finn Arup Nielsen, Finn Arup Nielsen

We describe a system for meta-analytical modeling of activation foci from functional neuroimaging studies. Our main vehicle is a set of density models in Talairach space capturing the distribution of...

Testing for difference between two groups of functional neuroimaging experiments (2007)

Pruning The Vocabulary For Better Context Recognition (2007)

We describe a meta-analytic method that tests for the difference between two groups of functional neuroimaging experiments. We use kernel density estimation in three-dimensional brain space to...

Rasmus Elsborg Madsen, Sigurdur Sigurdsson, Lars Kai Hansen, Jan Larsen

Abstract — Language independent ‘bag-of-words ’ representations are surprisingly effective for text classification. The representation is high dimensional though, containing many nonconsistent...

Bi-clique Communities (2007)

Lehmann, Sune, Schwartz, Martin, Hansen, Lars Kai

We present a novel method for detecting communities in bipartite networks. Based on an extension of the $k$-clique community detection algorithm, we demonstrate how modular structure in bipartite...

Deterministic Modularity Optimization (2007)

Lehmann, Sune, Hansen, Lars Kai

We study community structure of networks. We have developed a scheme for maximizing the modularity Q based on mean field methods. Further, we have defined a simple family of random networks with...

Sparse Kernel Orthonormalized PLS for feature extraction in large data sets (2007)

Jerónimo Arenas-garcía, Kaare Br, T Petersen, Lars Kai Hansen

In this paper we are presenting a novel multivariate analysis method. Our scheme is based on a novel kernel orthonormalized partial least squares (PLS) variant for feature extraction, imposing...

Sparse Kernel Orthonormalized PLS for feature extraction in large data sets (2007)

Jerónimo Arenas-garcía, Kaare Br, T Petersen, Lars Kai Hansen

In this paper we are presenting a novel multivariate analysis method. Our scheme is based on a novel kernel orthonormalized partial least squares (PLS) variant for feature extraction, imposing...

Unsupervised speaker change detection for broadcast news segmentation (2006)

Jørgensen, Kasper W., Mølgaard, Lasse, Hansen, Lars Kai

This paper presents a speaker change detection system for news broadcast segmentation based on a vector quantization (VQ) approach. The system does not make any assumption about the number of...

Learning and clean-up in a large scale music database (2006)

Hansen, Lars Kai, Lehn-Schiøler, Tue, Petersen, T, Arenas-Garcia, J, Larsen, Jan, Jensen, S

We have collected a database of musical features from radio broadcasts (N > 100.000). The database poses a number of hard modeling challenges including: Segmentation problems and missing metadata. We...

Temporal Feature Integration for Music Genre Classification (2006)

Meng, Anders, Ahrendt, Peter, Larsen, Jan, Hansen, Lars Kai

Temporal feature integration is the process of combining all the feature vectors in a time frame into a single feature vector in order to capture the relevant temporal information in the frame. The...

Adaptive regularization of noisy linear inverse problems (2006)

Hansen, Lars Kai, Lehn-Schiøler, Tue, Madsen, K. H

In the Bayesian modeling framework there is a close relation between regularization and the prior distribution over parameters. For prior distributions in the exponential family, we show that the...

Unsupervised speaker change detection for broadcast news segmentation (2006)

Kasper Jørgensen, Lasse Mølgaard, Lars Kai Hansen

This paper presents a speaker change detection system for news broadcast segmentation based on a vector quantization (VQ) approach. The system does not make any assumption about the number of...

The matrix cookbook (2006)

Kaare Brandt Petersen, Michael Syskind Pedersen, Suggestions Bill Baxter, Brian Templeton, Christian Rishøj, L. Theobald, ...

What is this? These pages are a collection of facts (identities, approximations, inequalities, relations,...) about matrices and matters relating to them. It is collected in this form for the...

Robust multi-scale clustering of large DNA microarray datasets with the consensus algorithm (2006)

Grotkjær, Thomas, Winther, Ole, Regenberg, Birgitte, Nielsen, Jens, Hansen, Lars Kai

Motivation: Hierarchical and relocation clustering (e.g. K-means and self-organizing maps) have been successful tools in the display and analysis of whole genome DNA microarray expression data....

Robust multi-scale clustering of large DNA microarray datasets with the consensus algorithm (2005)

Grotkjær, Thomas, Winther, Ole, Regenberg, Birgitte, Nielsen, Jens, Hansen, Lars Kai

Motivation: Hierarchical and relocation clustering (e.g. K-means and self-organising maps) have been successful tools in the display and analysis of whole genome DNA microarray expression data....

Part-of-speech enhanced context recognition (2004)

Madsen, Rasmus Elsborg, Larsen, Jan, Hansen, Lars Kai

Language independent `bag-of-words' representations are surprisingly efective for text classi¯cation. In this communi- cation our aim is to elucidate the synergy between language inde- pendent...

Enhanced context recognition by sensitivity pruned vocabularies (2004)

Madsen, Rasmus Elsborg, Sigurdsson, Sigurdur, Hansen, Lars Kai

Language independent `bag-of-words' representations are surprisingly effective for text classification. The generic BOW approach is based on a high-dimensional vocabulary which may reduce the...

Pruning the vocabulary for better context recognition (2004)

Madsen, Rasmus Elsborg, Sigurdsson, Sigurdur, Hansen, Lars Kai, Larsen, Jan

Language independent ‘bag-of-words’ representations are surprisingly effective for text classification. The representation is high dimensional though, containing many nonconsistent words for text...

Vocabulary pruning for improved context recognition (2004)

Madsen, Rasmus Elsborg, Sigurdsson, Sigurdur, Hansen, Lars Kai, Larsen, Jan

Language independent `bag-of-words' representations are surprisingly effective for text classification. The representation is high dimensional though, containing many non-consistent words for text...

Semi-blind Source Separation Using Head-Related Transfer Functions (2004)

Pedersen, Michael, Hansen, Lars Kai, Kjems, Ulrik, Rasmussen, Karsten Bo

An online blind source separation algorithm which is a special case of the geometric algorithm by Parra and Fancourt has been implemented for the purpose of separating sounds recorded at microphones...

Mapping from Speech to Images Using Continuous State Space Models (2004)

Lehn-Schiøler, Tue, Hansen, Lars Kai, Larsen, Jan

In this paper a system that transforms speech waveforms to animated faces are proposed. The system relies on continuous state space models to perform the mapping, this makes it possible to ensure...

Finn Arup Nielsen, Lars Kai Hansen

Identification of related research in functional neuroimaging can be done, e.g., by searching in bibliographic databases such as Pubmed, browsing "table of contents " of scientific...

Part-Of-Speech Enhanced Context (2004)

Recognition Rasmus Elsborg, Rasmus Elsborg Madsen, Jan Larsen, Lars Kai Hansen

Language independent `bag-of-words' representations are surprisingly e#ective for text classification. In this communication our aim is to elucidate the synergy between language independent...

Manual review (2004)

Finn ˚arup Nielsen, Daniela Balslev, Lars Kai Hansen

Abstract in Danish Functional brain imaging studier kortlgger funktion til hjerneomrder. Vi har anvendt automatisk usuperviseret textanalyse p PubMed databasen med henblik p at finde hvilke...

Estimating and suppressing background in Raman spectra with an artificial neural network. (2003)

Sigurdur Sigurdsson, Peter Alshede Philipsen, Monika Gniadecka, Hans Christian Wulf, Lars Kai Hansen

In this report we address the problem of skin fluorescence in feature extraction from Raman spectra of skin lesions. We apply a highly automated neural network method for suppressing skin...

Sensitivity Analysis (2003)

Sensitivity Analysis (2003)

Statements like ”this study demonstrates highly consistent findings ” or ”our results reveal a striking degree of overlap ” appear commonly in the literature. Such statements are typically...

Finn ˚arup Nielsen, Daniela Balslev, Lars Kai Hansen

We devise a general method for automatic metaanalyses in neuroscience and apply it on text data from published functional imaging studies to extract main functions associated with a brain area —...

Large scale bayesian kernel learning with the subspace algorithm (2002)

Time Series Prediction Based on the Relevance Vector Machine with Adaptive Kernels (2002)

Editor: unknown In this paper we consider Bayesian kernel learning and introduce a new way of using the expectation-maximization (EM) algorithm which amounts to taking the computationally most...

Joaquin Quiñonero-Candela, Lars Kai Hansen

The Relevance Vector Machine (RVM) introduced by Tipping is a probabilistic model similar to the widespread Support Vector Machines (SVM), but where the training takes place in a Bayesian framework,...

Automatic anatomical labeling of Talairach coordinates and generation of volumes of interest via the BrainMap database (2002)

An ICA algorithm for analyzing multiple data sets (2002)

BrainMap database

Ana S. Lukic, Miles N. Wernick, Lars Kai Hansen, Stephen C. Strother

In this paper we derive an independent-component analysis (ICA) method for analyzing two or more data sets simultaneously. Our model permits there to be components individual to the various data...

A spatially robust ica algorithm for multiple fMRI data sets (2002)

Ana S. Lukic, Miles N. Wernick, Lars Kai Hansen, Jon Anderson, Stephen C. Strother

In this paper we derive an independent-component analysis (ICA) method for analyzing two or more data sets simultaneously. Our model extracts independent components common to all data sets and...

Mean Field Implementation of Bayesian ICA (2001)

Pedro Højen-sørensen, Lars Kai Hansen, Mathematical Modelling

In this contribution we review the mean field approach to Bayesian independent component analysis (ICA) recently developed by the authors [1, 2]. For the chosen setting of additive Gaussian noise on...

Hierarchical clustering for datamining (2001)

Anna Szymkowiak, Jan Larsen, Lars Kai Hansen, Mathematical Modeling, Richard Petersens Plads

Abstract. This paper presents hierarchical probabilistic clustering methods for unsupervised and supervised learning in datamining applications. The probabilistic clustering is based on the...

Webmining: Learning from the World Wide Web (2001)

Jan Larsen, Lars Kai Hansen, Anna Szymkowiak, Torben Christiansen, Thomas Kolenda

Automated analysis of the world wide web is a new challenging area relevant in many applications, e.g., retrieval, navigation and organization of information, automated information assistants, and...

Neuroinformatics in Functional Neuroimaging (2001)

Finn ˚arup Nielsen, Lars Kai Hansen, Daniela Balslev

Mining for associations between text and brain activation in a

On Independent Component Analysis for Multimedia Signals (2000)

Jan Larsen, Lars Kai Hansen, Thomas Kolenda, Finn ˚arup Nielsen

Modeling of multimedia and multimodal data becomes increasingly important with the digitalization of the world. The objective of this paper is to demonstrate the potential of independent component...

On Independent Component Analysis for Multimedia Signals (2000)

Jan Larsen, Lars Kai Hansen, Thomas Kolenda, Finn ˚arup Nielsen

Modeling of multimedia and multimodal data becomes increasingly important with the digitalization of the world. The objective of this paper is to demonstrate the potential of independent component...

Webmining: learning from the world wide web (2000)

Jan Larsen, Lars Kai Hansen, Anna Szymkowiak, Torben Christiansen, Thomas Kolenda

Abstract: Automated analysis of the world wide web is a new challenging area relevant in many applications, e.g., retrieval, navigation and organization of information, automated information...

Bayesian averaging is well-temperated (2000)

Ensemble Learning and Linear Response Theory for (2000)

Bayesian predictions are stochastic just like predictions of any other inference scheme that generalize from a finite sample. While a simple variational argument shows that Bayes averaging is...

Ole Winther, Lars Kai Hansen

We propose a general Bayesian framework for performing independent component analysis (ICA) which relies on ensemble learning and linear response theory known from statistical physics. We apply it to...

Independent components in text (2000)

Thomas Kolenda, Lars Kai Hansen

In this communication we analyze the feasibility of independent component analysis (ICA) for dimensional reduction and representation of word histograms. The analysis is carried out in a likelihood...

Ensemble Learning and Linear Response Theory for ICA (2000)

Pedro Højen-Sørensen, Ole Winther, Lars Kai Hansen

We propose a general framework for performing independent component analysis (ICA) which relies on ensemble learning and linear response theory known from statistical physics. We apply it to both...

Experiences with Matlab and VRML in Functional Neuroimaging Visualizations (2000)

Finn Årup Nielsen, Lars Kai Hansen

Introduction Neuroinformatics is the task of organizing, analyzing and presenting the knowledge of neuroscience. In the part of neuroinformatics that is associated with functional neuroimaging...

Experiences with Matlab and VRML in Functional Neuroimaging Visualizations (2000)

Finn Arup Nielsen, Rup Nielsen, Lars Kai Hansen, Finn A, Finn A, Functional Neuroimaging, ...

so-called Talairach coordinates [Talairach and Tournoux, 1988]. 4 Finn A rup Nielsen and Lars Kai Hansen, DSP, IMM, DTU April 27, 2000 \LYNGBY" Figure 4: Lyngby Functional neuroimaging analysis...

On Comparison Of Adaptive Regularization Methods (2000)

Sigurdur Sigurdsson, Jan Larsen, Lars Kai Hansen

This paper investigates recently suggested adaptive regularization schemes.

Experiences with Matlab and VRML in Functional Neuroimaging Visualizations (2000)

Finn Årup Nielsen, Lars Kai Hansen

We describe some experiences with Matlab and VRML. We are developing a toolbox for neuroinformatics and describe some of the functionalities we have implemented or will implement and how Matlab and...

Modelling the fMRI response using smooth FIR filters (2000)

Finn Årup Nielsen, A. Nielsen, Lars Kai Hansen, Cyril Goutte, L. K. Hansen

Introduction We describe a flexible semi-parametric linear time-invariant convolution model for the fMRI response. It extends the FIR model [1] with a Gaussian process prior over the parameters (the...

Mining the BrainMap database: Detection of outliers (2000)

Finn Arup Nielsen, Lars Kai Hansen, Ulrik Kjems

We describe a system for modeling BrainMap -- a neuroscience database describing activation foci reported from many neuroimaging studies. We apply machine learning techniques in the form of...

On Independent Component Analysis for Multimedia Signals (2000)

Lars Kai Hansen, Jan Larsen, Thomas Kolenda

We discuss the independent component problem within a context of multimedia applications. The literature offers several independent component analysis schemes which can be applied in this context,...

Bayesian modelling of fMRI time series (2000)

Pedro Højen-Sørensen, Lars Kai Hansen, Carl Edward Rasmussen

We present a Hidden Markov Model (HMM) for inferring the hidden psychological state (or neural activity) during single trial fMRI activation experiments with blocked task paradigms. Inference is...

Modeling text with generalizable gaussian mixtures (2000)

Lars Kai Hansen, Sigurdur Sigurdsson, Thomas Kolenda, Finn Arup Nielsen, Ulrik Kjems, Jan Larsen

We apply and discuss generalizable Gaussian mixture (GGM) models for textmining. The model automatically adapts model complexity for a given text representation. We show that the generalizability of...

Bayesian averaging is well-temperated (2000)

Independent components in text (2000)

Bayesian predictions are stochastic just like predictions of any other inference scheme that generalize from a nite sample. While a simple variational argument shows that Bayes averaging is...

Thomas Kolenda, Lars Kai Hansen

In this communication we analyze the feasibility of independent component analysis (ICA) for dimensional reduction and representation of word histograms. The analysis is carried out in a likelihood...

On Independent Component Analysis for Multimedia Signals (2000)

Lars Kai Hansen, Jan Larsen, Thomas Kolenda

We discuss the independent component problem within a context of multimedia applications. The literature offers several independent component analysis schemes which can be applied in this context,...

Ensemble Learning and Linear Response Theory for (2000)

Ole Winther, Lars Kai Hansen

We propose a general framework for performing independent component analysis (ICA) which relies on ensemble learning and linear response theory known from statistical physics. We apply it to both...

On clustering fMRI time series (1999)

Peter Toft, Egill Rostrup, Finn A. Nielsen, Lars Kai Hansen, Cyril Goutte, Cyril Goutte

Running title: Clustering fMRI time series. Address for correspondence:

A Probabilistic Neural Network Framework for Detection of Malignant Melanoma (1999)

Mads Hintz-madsen, Lars Kai Hansen, Jan Larsen, Krzysztof T. Drzewiecki

Contents 1 INTRODUCTION 3 1.1 Malignant melanoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Evolution of malignant melanoma . . . . . . . . . . . . . . . . ....

Dynamic Components Of Linear Stable Mixtures From Fractional Low Order Moments (1999)

Thomas Fabricius Preben, Preben Kidmose, Lars Kai Hansen

The second moment based independent component analysis scheme of Molgedey and Schuster is generalized to fractional low order moments, relevant for linear mixtures of heavy tail stable processes. The...

Modeling Text With Generalizable Gaussian Mixtures (1999)

Lars Kai Hansen, Sigurdur Sigurdsson, Thomas Kolenda, Finn Arup Nielsen, Ulrik Kjems, Jan Larsen

We apply and discuss generalizable Gaussian mixture (GGM) models for textmining. The model automatically adapts model complexity for a given text representation. We show that the generalizability of...

Modeling Text With Generalizable Gaussian Mixtures (1999)

Lars Kai Hansen, Sigurdur Sigurdsson, Thomas Kolenda, Finn Arup Nielsen, Ulrik Kjems, Jan Larsen

We apply and discuss generalizable Gaussian mixture (GGM) models for textmining. The model automatically adapts model complexity for a given text representation. We show that the generalizability of...

Generalizable Patterns in Neuroimaging: How Many Principal Components? (1999)

Lars Kai Hansen, Jan Larsen, Finn Årup Nielsen, Arup Nielsen, Stephen C. Strother, Egill Rostrup, ...

Generalization can be defined quantitatively and can be used to assess the performance of Principal Component Analysis (PCA). The generalizability of PCA depends on the number of principal components...

Discrimination Of Cylinders With Different Wall Thicknesses Using Neural Networks And Simulated Dolphin Sonar Signals (1999)

Lars Nonboe Andersen, Whitlow Au, Jan Larsen, Lars Kai Hansen

This paper describes a method integrating neural networks into a system for recognizing underwater objects. The system is based on a combination of simulated dolphin sonar signals, simulated auditory...

Design of Robust Neural Network Classifiers (1998)

Jan Larsen, Lars Nonboe Andersen, Mads Hintz-madsen, Lars Kai Hansen

This paper addresses a new framework for designing robust neural network classifiers. The network is optimized using the maximum a posteriori technique, i.e., the cost function is the sum of the...

Source separation in short image sequences using delayed correlation (1998)

Neural Classifier Construction using Regularization, Pruning and Test Error Estimation (1998)

This paper addresses the use of blind sources separation techniques for analysis of short image sequences (multi-dimensional signals). We suggest a modification of Molgedey and Schuster's [8]...

Mads Hintz-madsen, Lars Kai Hansen, Jan Larsen, Morten Pedersen, Michael Larsen

In this paper we propose a method for construction of feed-forward neural classifiers based on regularization and adaptive architectures. Using a penalized maximum likelihood scheme, we derive a...

Biomedical Engineering at the Technical University of Denmark (1998)

Jørgen Arendt Jensen, Ole Trier Andersen, Jens E. Wilhjelm, Bent Kofoed, Lars Kai Hansen

The paper gives a brief overview of the biomedical engineering research and education at the Technical University of Denmark. An account of the research activities since the 1950s is given, and...

Blind Separation of Noisy Mixtures (1998)

Neuroinformatics based on VRML (1998)

We analyze blind separation of independent sources in the face of additive noise. The analysis is carried out in a likelihood framework, which allows estimates of the source signals, the mixing...

Finn Årup Nielsen, Lars Kai Hansen, F. A. Nielsen, L. K. Hansen

he figure we show an example of an environment summarizing and linking brain map data from a number of experiments in a "value added" Talairach frame of reference. Fiducial markers, grid...

Adaptive Regularization in Neural Network Modeling (1998)

Jan Larsen, Lars Nonboe Andersen, Lars Kai Hansen

Abstract. In this paper we address the important problem of optimizing regularization parameters in neural network modeling. The suggested optimization scheme is an extended version of the recently...

Generalization: The Hidden Agenda of Learning (1997)

Regularization with a Pruning Prior (1997)

Most neural systems are adapted by optimization of a performance index, typically the minimization of a \cost function", based on a nite database (a training set) of N noisy examples derived...

Cyril Goutte, Lars Kai Hansen

We investigate the use of a regularization prior that we show has pruning properties. Analyses are conducted both using a Bayesian framework and with the generalization method, on a simple toy...

Linear Unlearning for Cross-Validation (1996)

Unsupervised Learning and Generalization (1996)

The leave-one-out cross-validation scheme for generalization assessment of neural network models is computationally expensive due to replicated training sessions. In this paper we suggest linear...

Revisiting Boltzmann Learning: Parameter Estimation in Markov Random Fields (1996)

The concept of generalization is defined for a general class of unsupervised learning machines. The generalization error is a straightforward extension of the corresponding concept for supervised...

Lars Kai Hansen, Lars Nonboe Andersen, Ulrik Kjems, Jan Larsen

This contribution concerns a generalization of the Boltzmann Machine that allows us to use the learning rule for a much wider class of maximum likelihood and maximum a posteriori problems, including...

Cross-Validation with LULOO (1996)

Paul Haase Srensen, Magnus Nrgard, Lars Kai Hansen, Jan Larsen

Abstract--- The leave-one-out cross-validation scheme for generalization assessment of neural network models is computationally expensive due to replicated training sessions. Linear unlearning of...

Detection of malignant melanoma using neural classifiers (1996)

Mads Hintz-madsen, Lars Kai Hansen, Jan Larsen, Eric Olesen, Krzysztof T. Drzewiecki

In this paper we propose a method for design of feed-forward neural classifiers based on regularization and adaptive architectures, and we apply the scheme to the problem of detecting malignant...

Pruning with Generalization Based Weight Saliencies: flOBD, flOBS (1996)

Linear Unlearning for Cross-Validation (1996)

The purpose of most architecture optimization schemes is to improve generalization. In this presentation we suggest to estimate the weight saliency as the associated change in generalization error if...

Empirical Generalization Assessment of Neural Network Models (1995)

The leave-one-out cross-validation scheme for generalization assessment of neural network models is computationally expensive due to replicated training sessions. In this paper we suggest linear...

Design and Evaluation of Neural Classifiers Application to Skin Lesion Classification (1995)

Abstract. This paper addresses the assessment of generalization performance of neural network models by use of empirical techniques. We suggest to use the cross-validation scheme combined with a...

Mads Hintz-madsen, Lars Kai Hansen, Jan Larsen, Eric Olesen, Krzysztof T. Drzewiecki

We address design and evaluation of neural classifiers for the problem of skin lesion classification. By using Gauss Newton optimization for the entropic cost function in conjunction with pruning by...

The Error-Reject Tradeoff (1995)

Lars Kai Hansen, Christian Liisberg, Peter Salamon

We investigate the error versus reject tradeoff for classifiers. Our analysis is motivated by the remarkable similarity in error-reject tradeoff curves for widely differing algorithms classifying...

Fitness-functions of Genetic Algorithms for optimizing Neural Network topologies (1995)

Gunnar Øyro, Lars Kai Hansen

This paper presents a Genetic Algorithm (GA) designed for finding topologies of recurrent networks in order to optimize the performance in terms of learning error and generalization ability. In this...

Visualization of Neural Networks Using Saliency Maps (1995)

Ulrik Kjems, Lars Kai Hansen, Claus Svarer, Ian Law, ...

The saliency map is proposed as a new method for understanding and visualizing the nonlinearities embedded in feed-forward neural networks, with emphasis on the ill-posed case, where the...

Estimation of the Glucose Metabolism from Dynamic PET-scans using Neural Networks. (1995)

Claus Svarer, Ian Law, Søren Holm, Niels Mørch, Olaf Paulson, Lars Kai Hansen, ...

A method for fast pixel by pixel estimation of the Glucose Metabolism in the brain using the tracer [ 18 F]flourodeoxyglucose in dynamic PET-scan data is described. A neural network is trained to...

Regularized Parameter Estimation in an Inhomogeneous Cellular Network (1995)

Lars Nonboe Andersen, Lars Kai Hansen

The Maximum A Posteriori (MAP) approach has found ample use in signal processing [12, 4]. When applied to image data it leads to algorithms that map well onto networks of locally connected, simple...

Mean Field Reconstruction with Snaky Edge Hints (1995)

Peter Alshede Philipsen, Lars Kai Hansen, Peter Toft

Introduction Reconstruction of imagery of a non-ideal imaging system is a fundamental aim of computer vision. Geman and Geman [Geman(1984)] introduced Metropolis sampling from Gibbs distributions as...

Pruning from Adaptive Regularization (1994)

Lars Kai Hansen, Carl Edward Rasmussen

Inspired by the recent upsurge of interest in Bayesian methods we consider adaptive regularization. A generalization based scheme for adaptation of regularization parameters is introduced and...

Generalization performance of regularized neural network models (1994)