Median topographic maps for biomedical data sets (2009)
Hammer, Barbara, Hasenfuß, Alexander, Rossi, Fabrice
Median clustering extends popular neural data analysis methods such as the self-organizing map or neural gas to general data structures given by a dissimilarity matrix only. This offers flexible and...
09081 Summary -- Similarity-based learning on structures (2009)
Biehl, Michael, Hammer, Barbara, Hochreiter, Sepp, Kremer, Stefan C., Villmann, Thomas
The seminar centered around different aspects of similarity-based clustering with the special focus on structures. This included theoretical foundations, new algorithms, innovative applications, and...
09081 Abstracts Collection -- Similarity-based learning on structures (2009)
Biehl, Michael, Hammer, Barbara, Hochreiter, Sepp, Kremer, Stefan C., Villmann, Thomas
From 15.02. to 20.02.2009, the Dagstuhl Seminar 09081 ``Similarity-based learning on structures '' was held in Schloss Dagstuhl~--~Leibniz Center for Informatics. During the seminar, several...
Median topographic maps for biomedical data sets (2009)
Hammer, Barbara, Hasenfuß, Alexander, Rossi, Fabrice
Median clustering extends popular neural data analysis methods such as the self-organizing map or neural gas to general data structures given by a dissimilarity matrix only. This offers flexible and...
Median topographic maps for biomedical data sets (2009)
Hammer, Barbara, Hasenfuß, Alexander, Rossi, Fabrice
Median clustering extends popular neural data analysis methods such as the self-organizing map or neural gas to general data structures given by a dissimilarity matrix only. This offers flexible and...
Markovian bias of neural-based architectures with (2008)
Peter Tiňo, Barbara Hammer, Mikael Bodén
feedback connections
Neural Gas for Surface Reconstruction (2008)
Markus Melato, Barbara Hammer, Kai Hormann, Markus Melato, Barbara Hammer, Kai Hormann
In this paper we present an adaptation of Neural Gas (NG) for reconstructing 3Dsurfaces from point clouds. NG is based on online adaptation according to given data points and constitutes a neural...
IPK Gatersleben, Pattern Recognition Group, Gatersleben, (2008)
Barbara Hammer, Andreas Rechtien, Marc Strickert, Thomas Villmann
In medical classification tasks, it is important to gain in-
Marie Cottrell, Barbara Hammer, Er Hasenfuß, Thomas Villmann
Abstract- We introduce a batch variant of the neural gas (NG) clustering algorithm which optimizes the same cost function as NG but shows faster convergence. It has the additional benefit that, based...
Michael Biehl, Anarta Ghosh, Barbara Hammer
Learning vector quantization (LVQ) schemes constitute intuitive, powerful classification heuristics with numerous successful applications but, so far, limited theoretical background. We study LVQ...
Fabrice Rossi, Er Hasenfuß, Barbara Hammer
Abstract — In some application contexts, data are better described by a matrix of pairwise dissimilarities rather than by a vector representation. Clustering and topographic mapping algorithms have...
Matrix Learning in Learning Vector Quantization (2008)
Michael Biehl, Barbara Hammer, Petra Schneider, Michael Biehl, Barbara Hammer, Petra Schneider
We propose a new matrix learning scheme to extend Generalized Relevance Learning Vector Quantization (GRLVQ), an efficient prototype-based classification algorithm. By introducing a full matrix of...
Abstract Merge SOM for temporal data (2008)
Marc Strickert, Barbara Hammer
The recent merging self-organizing map (MSOM) for unsupervised sequence processing constitutes a fast, intuitive, and powerful unsupervised learning model. In this paper, we investigate its...
IPK Gatersleben, Pattern Recognition Group, Gatersleben, (2008)
Barbara Hammer, Andreas Rechtien, Marc Strickert, Thomas Villmann
In medical classification tasks, it is important to gain in-
Neural Gas for Surface Reconstruction (2008)
Markus Melato, Barbara Hammer, Kai Hormann
In this paper we present an adaptation of Neural Gas (NG) for reconstructing 3D-surfaces from point clouds. NG is based on online adaptation according to given data points and constitutes a neural...
with Correlation Measures for Gene Expression Analysis (2008)
Marc Strickert, Nese Sreenivasulu, Thomas Villmann, Barbara Hammer, Winfriede Weschke, Udo Seiffert
Abstract. A correlation-based similarity measure is derived for generalized relevance learning vector quantization (GRLVQ). The resulting classifier makes Pearson correlation available in a...
How to process uncertainty in machine learning? (2008)
Barbara Hammer, Thomas Villmann
Abstract. Uncertainty is a popular phenomenon in machine learning and a variety of methods to model uncertainty at different levels has been developed. The aim of this paper is to motivate the merits...
Relevance Determination in Reinforcement Learning (2008)
Abstract. We propose relevance determination and minimisation schemes in reinforcement learning which are solely based on the Q-matrix and which can thus be applied during training without prior...
Mapping the Design Space of Reinforcement Learning Problems – a Case Study (2008)
Barbara Hammer, Ag Lnm, Helge Ritter
This paper reports on a case study motivated by a typical reinforcement learning problem in robotics: an overall goal which decomposes into several subgoals has to be reached in a discrete large...
Pattern Recognition Group (2008)
Udo Seiffert, Barbara Hammer, Samuel Kaski, Thomas Villmann
Abstract. Bioinformatics is a promising and innovative research field. Despite of a high number of techniques specifically dedicated to bioinformatics problems as well as many successful...
Magnification control for batch neural gas (2008)
Barbara Hammer, Er Hasenfuß, Thomas Villmann
Abstract. It is well known, that online neural gas (NG) possesses a magnification exponent different from the information theoretically optimum one in adaptive map formation. The exponent can...
Klaus Ecker, Barbara Hammer, Klaus Ecker, Barbara Hammer (eds, J. Blazewicz, K. Ecker, ...
apl. Prof. Dr. Günter Kemnitz (Hardware and Robotics) Prof. Dr. Ingbert Kupka (Theoretical Computer Science) Prof. Dr. Wilfried Lex (Mathematical Foundations of Computer Science) Prof. Dr. Jörg...
Pattern Recognition Group (2008)
Udo Seiffert, Barbara Hammer, Samuel Kaski, Thomas Villmann
Abstract. Bioinformatics is a promising and innovative research field. Despite of a high number of techniques specifically dedicated to bioinformatics problems as well as many successful...
Abstract Unsupervised Recursive Sequence Processing (2008)
Marc Strickert, Barbara Hammer
The self organizing map (SOM) is a valuable tool for data visualization and data mining for potentially high dimensional data of an a priori fixed dimensionality. We investigate SOMs for sequences...
Supervised Median Clustering (2008)
Barbara Hammer, Er Hasenfuss, Frank-michael Schleif, Thomas Villmann, Barbara Hammer, Er Hasenfuss, ...
Prototype based clustering and classification algorithms constitute very intuitive and powerful machine learning tools for a variety of application areas. They combine simple training algorithms and...
Self-Organizing Maps for Time Series (2008)
Barbara Hammer Alessio, Barbara Hammer, Alessio Micheli, Nicolas Neubauer, Ro Sperduti, Marc Strickert
We review a recent extension of the self-organizing map (SOM) for temporal structures with a simple recurrent dynamics leading to sparse representations, which allows an e#cient training and a...
A general framework for unsupervised preocessing of structured data (2008)
Hammer, Barbara, Micheli, Alessio, Sperduti, Alessandro
We propose a general framework for unsupervised recurrent and recursive networks. This proposal covers various popular approaches like standard self organizing maps (SOM), temporal Kohonen maps,...
08041 Summary -- Recurrent Neural Networks - Models, Capacities, and Applications (2008)
De Raedt, Luc, Hammer, Barbara, Hitzler, Pascal, Maass, Wolfgang
The seminar centered around recurrent information processing in neural systems and its connections to brain sciences, on the one hand, and higher symbolic reasoning, on the other side. The goal was...
De Raedt, Luc, Hammer, Barbara, Hitzler, Pascal, Maass, Wolfgang
From January 20 to 25 2008, the Dagstuhl Seminar 08041 ``Recurrent Neural Networks- Models, Capacities, and Applications'' was held in the International Conference and Research Center (IBFI), Schloss...
Villmann, Thomas, Schleif, Frank-Michael, Kostrzewa, Markus, Walch, Axel, Hammer, Barbara
In the present contribution we propose two recently developed classification algorithms for the analysis of mass-spectrometric data—the supervised neural gas and the fuzzy-labeled self-organizing...
GLQ: A Journal of Lesbian and Gay Studies - Volume 14, Number 1, 2008
A NP-Hardness Result for a Sigmoidal 3-Node Neural Network (2007)
: In this paper we show that the loading problem for a fixed neural architecture with sigmoidal activation is NP-hard if the input dimension varies, if the classification is performed with a certain...
On the Generalization Capability of Simple Recurrent Neural Networks (2007)
: This paper establishes that simple Jordan and Elman networks with at least 4 resp. 5 hidden units have infinite Vapnik-Chervonenkis dimension. As a consequence the worst case generalization error...
Clinic for Psychotherapy and Psychosomatic Medicine, (2007)
Thomas Villmann, Barbara Hammer, Marc Strickert
In this contribution we combine approaches the generalized leraning vector quantization (GLVQ) with the neighborhood orientented learning in the neural gas network (NG). In this way we obtain a...
Barbara Hammer, Barbara Hammer, Alessio Micheli, Alessio Micheli, Ro Sperduti, Ro Sperduti
A general framework for self-organizing structure processing neural networks
B. Hammer, Compositionality in Neural Systems 1 Compositionality in Neural Systems (2007)
In real life, people deal with composite structures: Written English language is built of 26 characters and a few additional symbols which form syllables, words, sentences, articles, roadmaps, and...
2 The Resource Constrained Project Scheduling (2007)
Resource constraint project scheduling (RCPSP) is an NPhard benchmark problem in scheduling which takes into account the limitation of resources ' availabilities in real life production...
Vector Quantization with Rule Extraction for Mixed Domain Data (2007)
Barbara Hammer, Andreas Rechtien, Marc Strickert, Thomas Villmann
Abstract. We use a variant of learning vector quantization (LVQ) for extracting a rule based characterization of given labeled mixed domain data. Thereby, standard LVQ is improved to not only a more...
Thorsten Bojer, Barbara Hammer, Christian Koers
Abstract. We present an application of generalized relevance learning vector quantization (GRLVQ) to the supervision of piston compressors in industry. Thereby, GRLVQ constitutes a prototype-based...
Relevance Determination in Learning Vector Quantization (2007)
Thorsten Bojer, Barbara Hammer, Daniel Schunk
Abstract. We propose a method to automatically determine the relevance of the input dimensions of a learning vector quantization (LVQ) architecture during training. The method is based on Hebbian...
We examine the ability of combining symbolic and subsymbolic approaches by means of recursively encoding and decoding structured data. Symbolic data are tree structures { hence including formulas and...
Barbara Hammer, Thomas Villmann
Abstract. Recently a variation of learning vector quantization has been proposed in [1], which allows an automatic determination of relevance factors for the input dimensions: relevance learning...
Barbara Hammer, Thomas Villmann
Abstract. The neural gas algorithm provides a method to cluster a data space via an adaptive lattice of neurons which captures the topology of the data space. We propose dierent methods to determine...
Barbara Hammer, Alessio Micheli, Ro Sperduti
general framework for unsupervised processing of structured data
unsupervised processing (2007)
Barbara Hammer, Alessio Micheli, Marc Strickert, Alessandro Sperduti
A general framework for
Accelerating Relational Clustering Algorithms With Sparse Prototype Representation (2007)
Rossi, Fabrice, Hasenfuß, Alexander, Hammer, Barbara
In some application contexts, data are better described by a matrix of pairwise dissimilarities rather than by a vector representation. Clustering and topographic mapping algorithms have been adapted...
Advanced metric adaptation in Generalized LVQ for classification of mass spectrometry data (2007)
Schneider, Petra, Biehl, Michael, Schleif, Frank-Michael, Hammer, Barbara
Metric adaptation constitutes a powerful approach to improve the performance of prototype based classication schemes. We apply extensions of Generalized LVQ based on different adaptive distance...
Class imaging of hyperspectral satellite remote sensing data using FLSOM (2007)
Villmann, Thomas, Schleif, Frank-Michael, Merenyi, E., Strickert, M., Hammer, Barbara
We propose an extension of the self-organizing map for supervised fuzzy classification learning, whereby uncertain (fuzzy) class information is also allowed for training data. The method is able to...
Learning Vector Quantization: generalization ability and dynamics of competing prototypes (2007)
Witoelar, Aree, Biehl, Michael, Hammer, Barbara
Learning Vector Quantization (LVQ) are popular multi-class classification algorithms. Prototypes in an LVQ system represent the typical features of classes in the data. Frequently multiple prototypes...
Single pass clustering for large data sets (2007)
Alex, Nikolai, Hammer, Barbara, Klawonn, Frank
The presence of very large data sets poses new problems to standard neural clustering and visualization algorithms such as Neural Gas (NG) and the Self-Organizing-Map (SOM) due to memory and time...
Topographic Processing of Relational Data (2007)
Hammer, Barbara, Hasenfuß, Alexander, Rossi, Fabrice, Strickert, Marc
Recently, batch optimization schemes of the self-organizing map and neural gas have been modified to allow arbitrary distance measures.This principle is particularly suitable for complex applications...
Relational topographic maps (2007)
Barbara Hammer, Er Hasenfuss, Barbara Hammer, Er Hasenfuss
We introduce relational variants of neural topographic maps including the selforganizing map and neural gas, which allow clustering and visualization of data given in terms of a pairwise similarity...
We introduce relational variants of neural gas, a very efficient and powerful neural clustering algorithm. It is assumed that a similarity or dissimilarity matrix is given which stems from Euclidean...
Topographic processing of relational data (2007)
Barbara Hammer, Er Hasenfuss, Fabrice Rossi, Marc Strickert
Abstract — Recently, batch optimization schemes of the self-organizing map (SOM) and neural gas (NG) have been modified to so-called median variants which allow a transfer of these methods to...
Hammer, Barbara, Hasenfuss, Alexander
We introduce relational variants of neural gas, a very ecient and powerful neural clustering algorithm. It is assumed that a similarity or dissimilarity matrix is given which stems from Euclidean...
07131 Summary -- Similarity-based Clustering and its Application to Medicine and Biology (2007)
Biehl, Michael, Hammer, Barbara, Verleysen, Michel, Villmann, Thomas
This paper summarizes presentations, discussions, and results of the Dagstuhl seminar.
Biehl, Michael, Hammer, Barbara, Verleysen, Michel, Villmann, Thomas
From 25.03. to 30.03.2007, the Dagstuhl Seminar 07131 ``Similarity-based Clustering and its Application to Medicine and Biology'' was held in the International Conference and Research Center (IBFI),...
Learning Vector Quantization: generalization ability and dynamics of competing prototypes (2007)
Witoelar, Aree, Biehl, Michael, Hammer, Barbara
Learning Vector Quantization (LVQ) are popular multi-class classification algorithms. Prototypes in an LVQ system represent the typical features of classes in the data. Frequently multiple prototypes...
Batch and median neural gas (2006)
Cottrell, Marie, Hammer, Barbara, Hasenfuss, Alexander, Villmann, Thomas
Neural Gas (NG) constitutes a very robust clustering algorithm given euclidian data which does not suffer from the problem of local minima like simple vector quantization, or topological restrictions...
Batch and median neural gas (2006)
Cottrell, Marie, Hammer, Barbara, Hasenfuss, Alexander, Villmann, Thomas
Neural Gas (NG) constitutes a very robust clustering algorithm given euclidian data which does not suffer from the problem of local minima like simple vector quantization, or topological restrictions...
Batch and median neural gas (2006)
Cottrell, Marie, Hammer, Barbara, Hasenfuss, Alexander, Villmann, Thomas
Neural Gas (NG) constitutes a very robust clustering algorithm given euclidian data which does not suffer from the problem of local minima like simple vector quantization, or topological restrictions...
Neural networks and machine learning in bioinformatics - theory and applications (2006)
Seiffert, Udo, Hammer, Barbara, Kaski, Samuel, Villmann, Thomas
Bioinformatics is a promising and innovative research field. Despite of a high number of techniques specifically dedicated to bioinfor- matics problems as well as many successful applications, we are...
Batch and median neural gas (2006)
Cottrell, Marie, Hammer, Barbara, Hasenfuss, Alexander, Villmann, Thomas
Neural Gas (NG) constitutes a very robust clustering algorithm given euclidian data which does not suffer from the problem of local minima like simple vector quantization, or topological restrictions...
Batch and median neural gas (2006)
Cottrell, Marie, Hammer, Barbara, Hasenfuss, Alexander, Villmann, Thomas
Neural Gas (NG) constitutes a very robust clustering algorithm given euclidian data which does not suffer from the problem of local minima like simple vector quantization, or topological restrictions...
Dynamics and generalization ability of LVQ algorithms (2006)
Michael Biehl, Anarta Ghosh, Barbara Hammer, Yoshua Bengio
Learning vector quantization (LVQ) schemes constitute intuitive, powerful classification heuristics with numerous successful applications but, so far, limited theoretical background. We study LVQ...
Michael Biehl, Anarta Ghosh, Barbara Hammer
Winner-Takes-All (WTA) prescriptions for Learning Vector Quantization (LVQ) are studied in the framework of a model situation: Two competing prototype vectors are updated according to a sequence of...
Dynamical analysis of LVQ type learning rules (2005)
Anarta Ghosh, Michael Biehl, Barbara Hammer
Abstract- Learning vector quantization (LVQ) constitutes a powerful and simple method for adaptive nearest prototype classification which has been introduced based on heuristics. Recently, a...
Improving iterative repair strategies for scheduling with the SVM. Neurocomputing 63:271–292 (2005)
SVM
Barbara Hammer, Frank-michael Schleif, Thomas Villmann
We extend a recent variant of the prototype-based classifier learning vector quantization to a scheme which locally adapts relevance terms during learning. We derive explicit...
Improving iterative repair strategies for scheduling with the SVM. Neurocomputing 63:271–292 (2005)
Abstract. Resource constraint project scheduling (RCPSP) is an NPhard benchmark problem in scheduling which takes into account the limitation of resources ’ availabilities in real life production...
Classification using non standard metrics (2005)
Barbara Hammer, Thomas Villmann
A large variety of supervised or unsupervised learning algo-
Selforganizing maps for time series (2005)
Barbara Hammer, Alessio Micheli, Nicolas Neubauer, Ro Sperduti, Marc Strickert
Abstract- We review a recent extension of the self-organizing map (SOM) for temporal structures with a simple recurrent dynamics leading to sparse representations, which allows an efficient training...
The dynamics of learning vector quantization (2005)
Michael Biehl, Anarta Ghosh, Barbara Hammer
Abstract. Winner-Takes-All (WTA) algorithms offer intuitive and powerful learning schemes such as Learning Vector Quantization (LVQ) and variations thereof, most of which are heuristically motivated....
Relevance LVQ versus SVM (2004)
Barbara Hammer, Marc Strickert, Thomas Villmann
Abstract. The support vector machine (SVM) constitutes one of the most successful current learning algorithms with excellent classification accuracy in large real-life problems and strong theoretical...
Prototype based recognition of splice sites (2004)
Barbara Hammer, Marc Strickert, Thomas Villmann
Summary. Splice site recognition is an important subproblem of de novo gene finding, splice junctions constituting the boundary between coding and non-coding regions in eukaryotic DNA. The...
Universal Approximation Capability of Cascade Correlation Structures for (2004)
Barbara Hammer, Alessio Micheli
Cascade correlation (CC) constitutes a training method for neural networks which determines the weights as well as the neural architec-ture during training. Various extensions of CC to structured...
Prototype based recognition of splice sites (2004)
Barbara Hammer, Marc Strickert, Thomas Villmann
Summary. Splice site recognition is an important subproblem of de novo gene finding, splice junctions constituting the boundary between coding and non-coding regions in eukaryotic DNA. The...
Neural Methods for Non-Standard Data (2004)
Barbara Hammer, Brijnesh J. Jain
Standard pattern recognition provides effective and noise-tolerant tools for machine learning tasks; however, most approaches only deal with real vectors of a finite and fixed dimensionality. In this...
Neural methods for non-standard data (2004)
Barbara Hammer, Brijnesh J. Jain
Abstract. Standard pattern recognition provides effective and noise-tolerant tools for machine learning tasks; however, most approaches only deal with real vectors of a finite and fixed...
Universal Approximation Capability of Cascade Correlation for Structures (2004)
Barbara Hammer, Alessio Micheli, Ro Sperduti
Cascade correlation (CC) constitutes a training method for neural networks which determines the weights as well as the neural architec-ture during training. Various extensions of CC to structured...
Recurrent neural networks with small weights implement definite memory machines (2003)
definite memory machines
Neural maps in remote sensing image analysis (2003)
Thomas Villmann, Erzsébet Merényi, Barbara Hammer
We study the application of Self-Organizing Maps for the analyses of remote sensing spectral images. Advanced airborne and satellite-based imaging spectrometers produce very high-dimensional spectral...
Unsupervised recursive sequence processing (2003)
Marc Strickert, Barbara Hammer
Abstract. We propose a self organizing map (SOM) for sequences by extending standard SOM by two features, the recursive update of Sperduti [7] and the hyperbolic neighborhood of Ritter [5]. While the...
Perspectives on learning symbolic data with connectionistic systems (2003)
Abstract. This paper deals with the connection of symbolic and subsymbolic systems. It focuses on connectionistic systems processing symbolic data. We examine the capability of learning symbolic data...
Neural Gas for Sequences (2003)
Marc Strickert, Barbara Hammer
Keywords: sequence processing, temporal Kohonen map, recursive SOM, SOM for structured data, NG with context, HSOM Abstract — For unsupervised sequence processing, standard self organizing maps can...
Determining relevant input dimensions for the self-organizing map (2003)
Thorsten Bojer, Barbara Hammer, Marc Strickert, Thomas Villmann
Abstract. We propose a method to determine the relevance of the different input dimensions for a self organizing map (SOM). First, a growing self organizing map is adapted to the data. Afterwards,...
Neural Gas for Sequences (2003)
Marc Strickert, Barbara Hammer
Keywords: sequence processing, temporal Kohonen map, recursive SOM, SOM for structured data, NG with context, HSOM Abstract — For unsupervised sequence processing, standard self organizing maps...
Supervised Neural Gas with General Similarity Measure (2003)
Barbara Hammer, Marc Strickert, Thomas Villmann
Prototype based classi cation oers intuitive and sparse models with excellent generalization ability. However, these models usually crucially depend on the underlying Euclidian metric; moreover,...
Unsupervised recursive sequence processing (2003)
Marc Strickert, Barbara Hammer
Abstract. We propose a self organizing map (SOM) for sequences by extending standard SOM by two features, the recursive update of Sperduti [7] and the hyperbolic neighborhood of Ritter [5]. While the...
Markovian architectural bias of recurrent neural networks (2002)
We have recently shown that when initialized with "small " weights, recurrent neural networks (RNNs) with standard sigmoid-type activation functions are inherently biased towards...
Learning vector quantization for multimodal data (2002)
Barbara Hammer, Marc Strickert, Thomas Villmann
Abstract. Learning vector quantization (LVQ) as proposed by Kohonen is a simple and intuitive, though very successful prototype-based clustering algorithm. Generalized relevance LVQ (GRLVQ)...
Generalized relevance learning vector quantization (2002)
Barbara Hammer, Thomas Villmann
We propose a new scheme for enlarging generalized learning vector quantization (GLVQ) with weighting factors for the input dimensions. The factors allow an appropriate scaling of the input dimensions...
Rule extraction from self-organizing networks (2002)
Barbara Hammer, Andreas Rechtien, Marc Strickert, Thomas Villmann
Abstract. Generalized relevance learning vector quantization (GRLVQ) [4] constitutes a prototype based clustering algorithm based on LVQ [5] with energy function and adaptive metric. We propose a...
Markovian architectural bias of recurrent neural networks (2002)
We have recently shown that when initialized with “small ” weights, recur-rent neural networks (RNNs) with standard sigmoid-type activation functions are inherently biased towards Markov models,...
Recurrent networks for structured data - a unifying approach and its properties (2002)
its properties
Markovian architectural bias of recurrent neural networks (2002)
Abstract. We have recently shown that when initiated with \small" weights, recurrent neural networks (RNNs) with standard sigmoid-type activation functions are inherently biased towards...
Supervised neural gas for learning vector quantization (2002)
Thomas Villmann, Barbara Hammer, Marc Strickert
In this contribution we combine approaches the generalized leraning vector quantization (GLVQ) with the neighborhood oftentented learning in the neural gas network (NG). In this way we obtain a...
Rule extraction from self-organizing networks (2002)
Barbara Hammer, Andreas Rechtien, Marc Strickert, Thomas Villmann
Abstract. Generalized relevance learning vector quantization (GRLVQ) [4] constitutes a prototype based clustering algorithm based on LVQ [5] with energy function and adaptive metric. We propose a...
Learning vector quantization for multimodal data (2002)
Barbara Hammer, Marc Strickert, Thomas Villmann
Abstract. Learning vector quantization (LVQ) as proposed by Kohonen is a simple and intuitive, though very successful prototype-based clustering algorithm. Generalized relevance LVQ (GRLVQ)...
Markovian architectural bias of recurrent neural networks (2002)
We have recently shown that when initialized with \small " weights, recurrent neural networks (RNNs) with standard sigmoid-type activation functions are inherently biased towards Markov...
Generalized relevance LVQ for time series (2001)
Marc Strickert, Thorsten Bojer, Barbara Hammer
Abstract. An application of the recently proposed generalized relevance learning vector quantization (GRLVQ) to the analysis and modeling of time series data is presented. We use GRLVQ for two tasks:...
Estimating relevant input dimensions for self-organizing algorithms (2001)
Barbara Hammer, Thomas Villmann
Summary. We propose a new scheme for enlarging generalized learning vector quantization with weighting factors for the several input dimensions which are adapted according to the specific task. This...
Generalized relevance LVQ for time series (2001)
Marc Strickert, Thorsten Bojer, Barbara Hammer
Abstract. An application of the recently proposed generalized relevance learning vector quantization (GRLVQ) to the analysis and modeling of time series data is presented. We use GRLVQ for two tasks:...
On the generalization ability of recurrent networks (2001)
Abstract. The generalization ability of discrete time partially recurrent networks is examined. It is well known that the VC dimension of recurrent networks is infinite in most interesting cases and...
Generalization ability of folding networks (2001)
Abstract: The information theoretical learnability of folding networks, a very successful approach capable of dealing with tree structured inputs, is examined. We find bounds on the VC, pseudo-, and...
Self-Organizing Context Learning (2001)
Marc Strickert, Barbara Hammer
This work is designed to contribute to a deeper understanding of the recently proposed Merging SOM (MSOM). Its context model aims at the representation of sequences, an important subclass of...
Oracle Technology Network: Oracle Spatial UserÕs Guide and Reference (2001)
Marc Strickert, Barbara Hammer
Abstract. This work is designed to contribute to a deeper understanding of the recently proposed Merging SOM (MSOM). Its context model aims at the representation of sequences, an important subclass...
Thesis (doctoral)--Eberhard-Karls-Universität zu Tübingen, 2000.
On approximate learning by multi-layered feedforward circuits (2000)
Bhaskar Dasgupta, Barbara Hammer
We deal with the problem of efficient learning of feedforward neural networks. First, we consider the objective to maximize the ratio of correctly classified points compared to the size of the...
On approximate learning by multi-layered feedforward circuits (2000)
Bhaskar Dasgupta, Barbara Hammer
Abstract. We consider the problem of efficient approximate learning by multilayered feedforward circuits subject to two objective functions. First, we consider the objective to maximize the ratio of...
We consider the ability of neural encoding and decoding of symbolic structures which can be represented by tree structures and the correlated task of approximating and learning mappings from...
Limitations of hybrid systems (2000)
Abstract. We examine the ability of combining symbolic and subsymbolic approaches by means of recursively encoding and decoding structured data. We show that encoding of symbolic data is possible in...
Hardness of Approximation of the Loading Problem for Multi-layered Feedforward Neural Nets (2000)
Bhaskar Dasgupta, Barbara Hammer
We deal with the problem of e#cient learning of feedforward neural networks. First, we consider the objective to maximize the ratio of correctly classified points compared to the size of the training...
On Approximate Learning by Multi-layered Feedforward Circuits (Extended Abstract) (2000)
Bhaskar Dasgupta, Barbara Hammer
) Abstract We consider the problem of e#cient approximate learning by multi-layered feedforward circuits subject to two objective functions. First, we consider the objective to maximize the ratio of...
Neural Networks Classifying Symbolic Data (2000)
Introduction A very successful approach in machine learning are neural networks. They learn an unknown regularity between real vector spaces, given a finite set of examples. Hence they can be applied...
On approximate learning by multi-layered feedforward circuits (2000)
Bhaskar Dasgupta, Barbara Hammer
Abstract. We consider the problem of efficient approximate learning by multilayered feedforward circuits subject to two objective functions. First, we consider the objective to maximize the ratio of...
Thesis (M.A.)--Kent State University, 1999.
Approximation capabilities of folding networks (1999)
Abstract. In this paper we show several approximation results for folding networks-- a generalization of partial recurrent neural networks such that not only time sequences but arbitrary trees can...
On the approximation capability of recurrent neural networks (1999)
Abstract: The capability of recurrent neural networks of approximating functions from lists of real vectors to a real vector space is examined: Any measurable function can be approximated in...
On the approximation capability of recurrent neural networks (1999)
Abstract: The capability of recurrent neural networks to approximate functions from lists of real vectors to a real vector space is considered. We show the following results: From approximation...
Bhaskar Dasgupta, Barbara Hammer
In this paper we deal with the problem of ecient learning of feedforward neural networks. First, we consider the case when the objective is to maximize the ratio r of the correctly classied points...
On the learnability of recursive data (1999)
Abstract: We establish some general results concerning PAC learning: We find a characterization of the property that any consistent algorithm is PAC. It is shown that the shrinking width property is...
Learning with Recurrent Neural Networks (1999)
Barbara Hammer, Osnabruck Februar
: This thesis examines so-called folding neural networks as a mechanism for machine learning. Folding networks form a generalization of partial recurrent neural networks such that they are able to...
Hardness of Approximation of the Loading Problem for Multi-layered Feedforward Neural Nets (1999)
Bhaskar Dasgupta, Barbara Hammer
In this paper we deal with the problem of e#cient learning of feedforward neural networks. First, we consider the case when the objective is to maximize the ratio r of the correctly classified points...
Approximation capabilities of folding networks (1999)
Abstract. In this paper we show several approximation results for folding networks { a generalization of partial recurrent neural networks such that not only time sequences but arbitrary trees can...
An Interview with Barbara Hammer (1998)
Wide Angle - Volume 20, Number 1, January 1998
Training a sigmoidal network is difficult (1998)
In this paper we show that the loading problem for a 3-node architecture with sigmoidal activation is NP-hard if the input dimension varies, if the classification is performed with a certain...
Training a sigmoidal network is difficult (1998)
Abstract. In this paper we show that the loading problem for a 3-node architecture with sigmoidal activation is NP-hard if the input dimension varies, if the classification is performed with a...
Some complexity results for perceptron networks (1998)
The loading problem is the problem to decide if a neural architecture can map a training set correctly with an appropriate choice of the weights. The following results will be shown: The loading...
On the generalization of Elman networks (1997)
Abstract. The Vapnik Chervonenkis dimension of Elman networks is infinite. Here, we find constructions leading to lower bounds for the fat shattering dimension that are linear resp. of order log 2 in...
Learning Recursive Data is Intractable (1997)
We establish some general results concerning PAC learning: We find a characterization of the property, that any consistent algorithm is PAC. It is shown that the shrinking width property is...
Universal Approximation of Mappings on Structured Objects using the Folding Architecture (1996)
: The folding architecture is a universal mechanism to approximate mappings between trees and real vector spaces with a neural network. The part encoding the trees and the part approximating the...
Kiel, Univ., Diss., 1968.
Kiel, Med. F., Diss. v. 22. Febr. 1968.
On the Generalization Ability of GRLVQ networks
Barbara Hammer, Marc Strickert, Thomas Villmann
We derive a generalization bound for prototype-based classifiers with adaptive metric. The bound depends on the margin of the classifier and is independent of the dimensionality of the data. It holds...
On the Generalization Ability of GRLVQ Networks
Barbara Hammer, Marc Strickert, Thomas Villmann
We derive a generalization bound for prototype-based classifiers with adaptive metric. The bound depends on the margin of the classifier and is independent of the dimensionality of the data. It holds...