Transfer Learning by Distribution Matching for Targeted Advertising (2009)
Steffen Bickel, Christoph Sawade, Tobias Scheffer
We address the problem of learning classifiers for several related tasks that may differ in their joint distribution of input and output variables. For each task, small – possibly even empty –...
Transfer Learning by Distribution Matching for Targeted Advertising (2009)
Steffen Bickel, Christoph Sawade, Tobias Scheffer
We address the problem of learning classifiers for several related tasks that may differ in their joint distribution of input and output variables. For each task, small – possibly even empty –...
Multi-Task Learning for HIV Therapy Screening (2009)
Steffen Bickel, Jasmina Bogojeska, Thomas Lengauer, Tobias Scheffer
We address the problem of learning classifiers for a large number of tasks. We derive a solution that produces resampling weights which match the pool of all examples to the target distribution of...
Exact and Approximate Inference for Annotating Graphs with Structural SVMs (2008)
Klein, Thoralf, Brefeld, Ulf, Scheffer, Tobias
Training processes of structured prediction models such as structural SVMs involve frequent computations of the maximum-a-posteriori (MAP) prediction given a parameterized model. For specific output...
Discriminative Identification of Duplicates (2008)
Peter Haider, Ulf Brefeld, Tobias Scheffer
Abstract. The problem of finding duplicates in data is ubiquitous in data mining. We cast the problem of finding duplicates in sequential data into a poly-cut problem on a fully connected graph. The...
Predicting Sentences using N-Gram Language Models (2008)
Steffen Bickel, Peter Haider, Tobias Scheffer
We explore the benefit that users in several application areas can experience from a “tab-complete ” editing assistance function. We develop an evaluation metric and adapt N-gram language models...
Highly Scalable Discriminative Spam Filtering (2008)
Michael Brückner, Peter Haider, Tobias Scheffer
This paper discusses several lessons learned from the SpamTREC 2006 challenge. We discuss issues related to decoding, preprocessing, and tokenization of email messages. Using the Winnow algorithm...
Supervised Clustering for Spam Detection in Data Streams (2008)
Ulf Brefeld, Peter Haider, Tobias Scheffer
Introduction. We address the problem of detecting batches of emails that have been created according to the same template. This problem is motivated by the desire to filter spam more effectively by...
Efficiency and Stability of Clustering Algorithms for Linked Data (2008)
We are interested in finding clusters (“communities”) in networks of linked data, such as citation networks or web pages. Hierarchical clustering for networks is reviewed and an algorithmic...
Multi-view Learning and Link Farm Discovery (2008)
Abstract. The first part of this abstract focuses on estimation of mixture models for problems in which multiple views of the instances are available. Examples of this setting include clustering web...
Supervised Clustering of Streaming Data for Email Batch Detection (2008)
Peter Haider, Ulf Brefeld, Tobias Scheffer
We address the problem of detecting batches of emails that have been created according to the same template. This problem is motivated by the desire to filter spam more effectively by exploiting...
Predicting Sentences using N-Gram Language Models ∗ (2008)
Steffen Bickel, Peter Haider, Tobias Scheffer
We explore the benefit that users in several application areas can experience from a “tab-complete ” editing assistance function. We develop an evaluation metric and adapt N-gram language models...
Vorsitzender Prof, Dr. Stefan Jähnichen, Berichter Prof, Dr. Fritz Wysotzki, Berichter Prof, Dr. Claus Weihs, ...
I wish to thank everyone who helped me with my studies and contributed to my thesis. In particular, I wish to thank Eckehardt Blanz who supported my attempt to obtain the Ernst von Siemens fellowship...
Accurate Schema Matching on Streams ⋆ (2008)
Szymon Jaroszewicz, Lenka Ivantysynova, Tobias Scheffer
Abstract. We address the problem of matching imperfectly documented schemas of data streams and large databases. Instance-level schema matching algorithms identify likely correspondences between...
Discriminative Identification of Duplicates (2008)
Peter Haider, Ulf Brefeld, Tobias Scheffer
Abstract. The problem of finding duplicates in data is ubiquitous in data mining. We cast the problem of finding duplicates in sequential data into a poly-cut problem on a fully connected graph. The...
Multi-View Hidden Markov Perceptrons (2008)
Ulf Brefeld, Christoph Büscher, Tobias Scheffer
Discriminative learning techniques for sequential data have proven to be more effective than generative models for named entity recognition, information extraction, and other tasks of discrimination.
Multi-Task Learning for HIV Therapy Screening (2008)
Bickel, Steffen, Bogojeska, Jasmina, Lengauer, Thomas, Scheffer, Tobias
We address the problem of learning classifers for a large number of tasks. We derive a solution that produces resampling weights which match the pool of all examples to the target distribution of any...
Combination of Problem Solving and Learning from Experience Extended Abstract (2007)
Linda Briesemeister, Barbara Van Schewick, Tobias Scheffer
Most AI problem-solving systems, when presented with the same problem repeatedly, always solve it the same way and in about the same amount of time. It seems shortsighted that they do not adjust...
Graph Based Subsumption Algorithms for Machine Learning (2007)
Tobias Scheffer, Ralf Herbrich, Fritz Wysotzki
The `-subsumption problem is crucial to the efficiency of ILP learning systems. We discuss a graph-based `-subsumption algorithm for preselecting suitable matching literals. We further map the...
Combination of Problem Solving and Learning from Experience Extended Abstract (2007)
Linda Briesemeister, Barbara Van Schewick, Tobias Scheffer
Most AI problem-solving systems, when presented with the same problem repeatedly, always solve it the same
Unbiased Assessment of Learning Algorithms (Extended Abstract) (2007)
Tobias Scheffer, Ralf Herbrich
In order to rank the performance of machine learning algorithms, many researchers conduct experiments on benchmark data sets. Since most learning algorithms have domain-specific parameters, it is a...
ERROR ESTIMATION AND MODEL SELECTION vorgelegt von Dipl.-Inform. (2007)
Tobias Scheffer, Doktor Der Naturwissenschaften, Vorsitzender Prof, Dr. Stefan Jahnichen, Berichter Prof, ...
I wish to thank everyone who helped me with my studies and contributed to my thesis. In particular, I wish to thank Eckehardt Blanz who supported my attempt to obtain the Ernst von Siemens fellowship...
Many individuals, organizations, and companies have to answer large amounts of emails. Often, many of these emails contain variations of relatively few frequently asked questions. We address the...
We developed a well-structured term-based language for structural specification of artificial neural networks. The language achieves an intuitive and compact representation even for very large...
A Term-Based Genetic Code for Artificial Neural Networks (2007)
We developed a well-structured term-based language for the structural specification of artificial neural networks. The language achieves an intuitive and compact representation even for very large...
Efficient Algorithms for `-Subsumption (2007)
Tobias Scheffer, Ralf Herbrich, Fritz Wysotzki
` subsumption is a decidable but incomplete approximation of logic implication, important to inductive logic programming and theorem proving. We show that by context based elimination of possible...
Tobias Scheffer, Thorsten Joachims
Abstract. Model selection is considered the problem of choosing a "good " hypothesis language from a given ensemble of models. Here, a "good " model is one for...
Dirichlet-enhanced spam filtering based on biased samples (2007)
Steffen Bickel, Tobias Scheffer
We study a setting that is motivated by the problem of filtering spam messages for many users. Each user receives messages according to an individual, unknown distribution, reflected only in the...
Dirichlet-enhanced spam filtering based on biased samples (2007)
Steffen Bickel, Tobias Scheffer
We study a setting that is motivated by the problem of filtering spam messages for many users. Each user receives messages according to an individual, unknown distribution, reflected only in the...
Unsupervised prediction of citation influences (2007)
Laura Dietz, Steffen Bickel, Tobias Scheffer
Publication repositories contain an abundance of information about the evolution of scientific research areas. We address the problem of creating a visualization of a research area that describes the...
Discriminative learning for differing training and test distributions (2007)
Steffen Bickel, Michael Brückner, Tobias Scheffer
We address classification problems for which the training instances are governed by a distribution that is allowed to differ arbitrarily from the test distribution—problems also referred to as...
Transductive support vector machines for structured variables (2007)
Alexander Zien, Ulf Brefeld, Tobias Scheffer
We study the problem of learning kernel machines transductively for structured output variables. Transductive learning can be reduced to combinatorial optimization problems over all possible...
Unsupervised Prediction of Citation Influences (2007)
Dietz, Laura, Bickel, Steffen, Scheffer, Tobias
Abstract Publication repositories contain an abundance of information about the evolution of scientific research areas. We address the problem of creating a visualization of a research area that...
Semi-supervised learning for structured output variables (2006)
The problem of learning a mapping between input and structured, interdependent output variables covers sequential, spatial, and relational learning as well as predicting recursive structures. Joint...
Semi-supervised learning for structured output variables (2006)
The problem of learning a mapping between input and structured, interdependent output variables covers sequential, spatial, and relational learning as well as predicting recursive structures. Joint...
Multi-View Learning and Link Farm Discovery (2006)
The first part of this abstract focuses on estimation of mixture models for problems in which multiple views of the instances are available. Examples of this setting include clustering web pages or...
Transductive Learning for Text Classification using Explicit Knowledge Models (2006)
Ifrim, Georgiana, Weikum, Gerhard, Fürnkranz, Johannes, Scheffer, Tobias, Spiliopoulou, Myra
We present a generative model based approach for transductive learning for text classification. Our approach combines three methodological ingredients: learning from background corpora, latent...
Multiview discriminative sequential learning (2005)
1 Introduction The problem of labeling observation sequences has applications that range fromlanguage processing tasks such as named entity recognition, part-of-speech tagging, and information...
Discovering communities in linked data by multi-view clustering (2005)
Isabel Drost, Steffen Bickel, Tobias Scheffer
Abstract. We consider the problem of finding communities in large linked networks such as web structures or citation networks. We review similarity measures for linked objects and discuss the k-Means...
Estimation of mixture models using Co-EM (2005)
Steffen Bickel, Tobias Scheffer
We study estimation of mixture models for problems in which multiple views of the instances are available. Examples of this setting include clustering web pages or research papers that have intrinsic...
AUC Maximizing Support Vector Learning (2005)
The area under the ROC curve (AUC) is a natural performance measure when the goal is to find a discriminative decision function.
Learning to complete sentences (2005)
Steffen Bickel, Peter Haider, Tobias Scheffer
Abstract. We consider the problem of predicting how a user will continue a given initial text fragment. Intuitively, our goal is to develop a “tab-complete ” function for natural language, based...
Report Systematic feature evaluation for gene name recognition (2005)
Bmc Bioinformatics, Jörg Hakenberg, Steffen Bickel, Conrad Plake, Ulf Brefeld, Hagen Zahn, ...
In task 1A of the BioCreAtIvE evaluation, systems had to be devised that recognize words and phrases forming gene or protein names in natural language sentences. We approach this problem by building...
Multiview discriminative sequential learning (2005)
Ulf Brefeld, Christoph Büscher, Tobias Scheffer
Abstract. Discriminative learning techniques for sequential data have proven to be more effective than generative models for named entity recognition, information extraction, and other tasks of...
Estimation of mixture models using Co-EM (2005)
Steffen Bickel, Tobias Scheffer
Abstract. We study estimation of mixture models for problems in which multiple views of the instances are available. Examples of this setting include clustering web pages or research papers that have...
Thwarting the nigritude ultramarine: learning to identify link spam (2005)
Abstract. The page rank of a commercial web site has an enormous economic impact because it directly influences the number of potential customers that find the site as a highly ranked search engine...
Estimation of mixture models using Co-EM (2005)
Steffen Bickel, Tobias Scheffer
We study estimation of mixture models for problems in which multiple views of the instances are available. Examples of this setting include clustering web pages or research papers that have intrinsic...
Co-em support vector learning (2004)
Multi-view algorithms, such as co-training and co-EM, utilize unlabeled data when the available attributes can be split into independent and compatible subsets. Co-EM outperforms co-training for many...
Co-em support vector learning (2004)
Multi-view algorithms, such as co-training and co-EM, utilize unlabeled data when the available attributes can be split into independent and compatible subsets. Co-EM outperforms co-training for many...
Learning from Message Pairs for Automatic Email Answering (2004)
Steffen Bickel, Tobias Scheffer
We consider the problem of learning a mapping from question to answer messages. The training data for this problem consist of pairs of messages that have been received and sent in the past. We...
Multi-relational learning, text mining, and semi-supervised learning for functional genomics (2004)
Abstract. We focus on the problem of predicting functional properties of the proteins corresponding to genes in the yeast genome. Our goal is to study the effectiveness of approaches that utilize all...
A Support Vector Machine classifier for gene name recognition (2004)
Steffen Bickel, Ulf Brefeld, Lukas Faulstich, Jörg Hakenberg, Ulf Leser, Conrad Plake, ...
This summary describes our solution for task 1A of the BioCreAtIvE Challenge Cup 2003. Essentially, we reduce the entity recognition problem to the problem of classifying single words using a Support...
Combining data and text mining techniques for yeast gene regulation prediction: A case study (2003)
Mark-a. Krogel, Marcus Denecke, Marco L, Tobias Scheffer
Abstract. We focus on the problem of predicting yeast gene regulation experiments. In order to construct a good solution, we study combinations of different methods that are not yet to be found in...
• Luc Dehaspe, PharmaDM: “Great Expectations: A To-Do List for the Biologist’s in Silico (2003)
Tobias Scheffer, Ulf Leser (editors
In the past years, research in molecular biology and molecular medicine has accumulated enormous amounts of data. This includes genomic sequences gathered by the Human Genome Project, gene expression...
Combining Data and Text Mining Techniques for Yeast Gene Regulation Prediction: A Case Study (2003)
Mark-A. Krogel, Marcus Denecke, Marco Landwehr, Tobias Scheffer
In order to solve task 2 of the KDD Cup 2002, we exploited various available information sources. In particular, use of relational information describing the interactions among genes and information...
Finding the Most Interesting Patterns in a Database Quickly by Using Sequential Sampling (2002)
Tobias Scheffer, Fhg Ais, E. Brodley, Andrea Danyluk
Many discovery problems, e.g., subgroup or association rule discovery, can naturally be cast as nbest hypotheses problems where the goal is to find the n hypotheses from a given hypothesis space that...
Finding the Most Interesting Patterns in a Database Quickly by Using Sequential Sampling (2002)
Tobias Scheffer, Stefan Wrobel, E. Brodley, Andrea Danyluk
Many discovery problems, e.g., subgroup or association rule discovery, can naturally be cast as n- best hypotheses problems where the goal is to find the n hypotheses from a given hypothesis space...
Mining the web with active hidden markov models (2001)
Tobias Scheffer, Stefan Wrobel
Given the enormous amounts of information available only in unstructured or semi-structured textual documents, tools for information extraction (IE) have become enormously
Learning Hidden Markov Models for Information Extraction Actively from Partially Labeled Text (2001)
Tobias Scheffer, Stefan Wrobel, Borislav Popov, Damyan Ognianov, Christian Decomain, Susanne Hoche
A vast range of information is expressed in unstructured or semi-structured text, in a form that is hard to decipher automatically. Consequently, it is of enormous importance to construct tools that...
Finding Association Rules that Trade Support Optimally Against Confidence (2001)
When evaluating association rules, rules that differ in both support and confidence have to be compared; a larger support has to be traded against a higher confidence. The solution which we propose...
A sequential sampling algorithm for a general class of utility criteria (2000)
Tobias Scheffer, Stefan Wrobel
Many discovery problems, e.g., subgroup or association rule discovery, can naturally be cast as n-best hypothesis problems where the goal is to nd the n hypotheses from a given hypothesis space that...
Nonparametric Regularization of Decision Trees (2000)
We discuss the problem of choosing the complexity of a decision tree (measured in the number of leaf nodes) that gives us highest generalization performance. We first discuss an analysis of the...
Expected error analysis for model selection (1999)
Tobias Scheffer, Thorsten Joachims
In order to select a good hypothesis language (or model) from a collection of possible models, one has to assess the generalization performance of the hypothesis which is returned by a learner that...
Expected error analysis for model selection (1999)
Tobias Scheffer, Thorsten Joachims
In order to select a good hypothesis language (or model) from a collection of possible models, one has to assess the generalization performance of the hypothesis which is returned by a learner that...
Andrew Mitchell, Tobias Scheffer, Arun Sharma, Frank Stephan
Abstract. This paper derives the Vapnik Chervonenkis dimension of several natural subclasses of pattern languages. For classes with unbounded VC-dimension, an attempt is made to quantify the...
Error Estimation and Model Selection (1999)
In order to select a good hypothesis language (or model class) from a collection of possible model classes, we have to assess the generalization performance of the hypothesis which is returned by a...
Expected error analysis for model selection (1999)
Tobias Scheffer, Thorsten Joachims
In order to select a good hypothesis language (or model) from a collection of possible models, one has to assess the generalization performance of the hypothesis which is returned by a learner that...
Estimating the expected error of empirical minimizers for model selection (1998)
Tobias Scheffer, Thorsten Joachims
Model selection [e.g., 1] is considered the problem of choosing a hypothesis language which provides an optimal balance between low empirical error and high structural complexity. In this Abstract,...
Generation of Task-Specific Segmentation Procedures as a Model Selection Task (1998)
Ralf Herbrich, Tobias Scheffer
In image segmentation problems, there is usually a vast amount of filter operations available, a subset of which has to be selected and instantiated in order to obtain a satisfactory segmentation...
Discovering Association Rules with High Predictive Power (1998)
. I argue that association rules are only useful if they express regularities in the process that created a particular database, rather than regularities in the database itself -- which may be due to...
Unbiased assessment of learning algorithms (1997)
Tobias Scheffer, Ralf Herbrich
In order to rank the performance of machine learning algorithms, many researchers conduct experiments on benchmark data sets. Since most learning algorithms have domain-specific parameters, it is a...
Why experimentation can be better than perfect guidance (1997)
Tobias Scheffer, Russell Greiner, Christian Darken
The full version of this paper appeared at ICML-97. Many problems correspond to the classical control task of determining the appropriate control action to take, given some (sequence of)...
Generation of Task-Specific Segmentation Procedures as a Model Selection Task (1997)
Ralf Herbrich, Tobias Scheffer
In image segmentation problems, there is usually a vast amount of filter operations available, a subset of which has to be selected and instantiated in order to obtain a satisfactory segmentation...
Why Experimentation can be better than "Perfect Guidance" (1997)
Tobias Scheffer, Russell Greiner, Christian Darken
Many problems correspond to the classical control task of determining the appropriate control action to take, given some (sequence of) observations. One standard approach to learning these control...
Why Experimentation can be better than "Perfect Guidance" (1997)
Tobias Scheffer, Russell Greiner, Christian Darken
Many problems correspond to the classical control task of determining the appropriate control action to take, given some (sequence of) observations. One standard approach to learning these control...
A Concept Formation Based Algorithmic Model for Skill Acquisition (1996)
Linda Briesemeister, Tobias Scheffer
We present an algorithmic model for acquisition of cognitive skills that is based on machine learning and problem solving algorithms. The principle is to use a problem solving approach for new...
Algebraic foundation and improved methods of induction of ripple down rules (1996)
Ripple down rules (RDR), that is rules with hierarchical exceptions, are used in knowledge acquisition because they provide a well intelligible and modifiable representation for even very large...
Efficient -subsumption based on graph algorithms (1996)
Tobias Scheffer, Ralf Herbrich, Fritz Wysotzki
Abstract. The `-subsumption problem is crucial to the efficiency of ILP learning systems. We discuss two `-subsumption algorithms based on strategies for preselecting suitable matching literals. The...