Volker Tresp

Learning Gaussian processes from multiple tasks

We consider the problem of multi-task learning, that is, learning multiple related functions. Our approach is based on a hierarchical Bayesian framework, that exploits the equivalence between parametric linear models and nonparametric Gaussian processes (GPs). The resulting models can be learned easily via an EM-algorithm. Empirical studies on multi-label text categorization suggest that the presented models allow accurate solutions of these multi-task problems.

A Bayesian Committee Machine

The Bayesian committee machine (BCM) is a novel approach to combining estimators that were trained on different data sets. Although the BCM can be applied to the combination of any kind of estimators, the main foci are gaussian process regression and related systems such as regularization networks and smoothing splines for which the degrees of freedom increase with the number of training data. Somewhat surprisingly, we find that the performance of the BCM improves if several test points are queried at the same time and is optimal if the number of test points is at least as large as the degrees of freedom of the estimator. The BCM also provides a new solution for on-line learning with potential applications to data mining. We apply the BCM to systems with fixed basis functions and discuss its relationship to gaussian process regression. Finally, we show how the ideas behind the BCM can be applied in a non-Bayesian setting to extend the input-dependent combination of estimators.

Active learning via transductive experimental design

This paper considers the problem of selecting the most informative experiments x to get measurements y for learning a regression model y = f(x). We propose a novel and simple concept for active learning, transductive experimental design, that explores available unmeasured experiments (i.e., unlabeled data) and has a better scalability in comparison with classic experimental design methods. Our in-depth analysis shows that the new method tends to favor experiments that are on the one side hard-to-predict and on the other side representative for the rest of the experiments. Efficient optimization of the new design problem is achieved through alternating optimization and sequential greedy search. Extensive experimental results on synthetic problems and three real-world tasks, including questionnaire design for preference learning, active learning for text categorization, and spatial sensor placement, highlight the advantages of the proposed approaches.

Factorizing YAGO: scalable machine learning for linked data

Vast amounts of structured information have been published in the Semantic Webs Linked Open Data (LOD) cloud and their size is still growing rapidly. Yet, access to this information via reasoning and querying is sometimes difficult, due to LODs size, partial data inconsistencies and inherent noisiness. Machine Learning offers an alternative approach to exploiting LODs data with the advantages that Machine Learning algorithms are typically robust to both noise and data inconsistencies and are able to efficiently utilize non-deterministic dependencies in the data. From a Machine Learning point of view, LOD is challenging due to its relational nature and its scale. Here, we present an efficient approach to relational learning on LOD data, based on the factorization of a sparse tensor that scales to data consisting of millions of entities, hundreds of relations and billions of known facts. Furthermore, we show how ontological knowledge can be incorporated in the factorization to improve learning results and how computation can be distributed across multiple nodes. We demonstrate that our approach is able to factorize the YAGO~2 core ontology and globally predict statements for this large knowledge base using a single dual-core desktop computer. Furthermore, we show experimentally that our approach achieves good results in several relational learning tasks that are relevant to Linked Data. Once a factorization has been computed, our model is able to predict efficiently, and without any additional training, the likelihood of any of the 4.3 ⋅ 1014 possible triples in the YAGO~2 core ontology.

A Review of Relational Machine Learning for Knowledge Graphs

Relational machine learning studies methods for the statistical analysis of relational, or graph-structured, data. In this paper, we provide a review of how such statistical models can be “trained” on large knowledge graphs, and then used to predict new facts about the world (which is equivalent to predicting new edges in the graph). In particular, we discuss two fundamentally different kinds of statistical relational models, both of which can scale to massive data sets. The first is based on latent feature models such as tensor factorization and multiway neural networks. The second is based on mining observable patterns in the graph. We also show how to combine these latent and observable models to get improved modeling power at decreased computational cost. Finally, we discuss how such statistical models of graphs can be combined with text-based information extraction methods for automatically constructing knowledge graphs from the Web. To this end, we also discuss Googles knowledge vault project as an example of such combination.

Extraction of semantic biomedical relations from text using conditional random fields

BackgroundThe increasing amount of published literature in biomedicine represents an immense source of knowledge, which can only efficiently be accessed by a new generation of automated information extraction tools. Named entity recognition of well-defined objects, such as genes or proteins, has achieved a sufficient level of maturity such that it can form the basis for the next step: the extraction of relations that exist between the recognized entities. Whereas most early work focused on the mere detection of relations, the classification of the type of relation is also of great importance and this is the focus of this work. In this paper we describe an approach that extracts both the existence of a relation and its type. Our work is based on Conditional Random Fields, which have been applied with much success to the task of named entity recognition.ResultsWe benchmark our approach on two different tasks. The first task is the identification of semantic relations between diseases and treatments. The available data set consists of manually annotated PubMed abstracts. The second task is the identification of relations between genes and diseases from a set of concise phrases, so-called GeneRIF (Gene Reference Into Function) phrases. In our experimental setting, we do not assume that the entities are given, as is often the case in previous relation extraction work. Rather the extraction of the entities is solved as a subproblem. Compared with other state-of-the-art approaches, we achieve very competitive results on both data sets. To demonstrate the scalability of our solution, we apply our approach to the complete human GeneRIF database. The resulting gene-disease network contains 34758 semantic associations between 4939 genes and 1745 diseases. The gene-disease network is publicly available as a machine-readable RDF graph.ConclusionWe extend the framework of Conditional Random Fields towards the annotation of semantic relations from text and apply it to the biomedical domain. Our approach is based on a rich set of textual features and achieves a performance that is competitive to leading approaches. The model is quite general and can be extended to handle arbitrary biological entities and relation types. The resulting gene-disease network shows that the GeneRIF database provides a rich knowledge source for text mining. Current work is focused on improving the accuracy of detection of entities as well as entity boundaries, which will also greatly improve the relation extraction performance.

Supervised probabilistic principal component analysis

Principal component analysis (PCA) has been extensively applied in data mining, pattern recognition and information retrieval for unsupervised dimensionality reduction. When labels of data are available, e.g., in a classification or regression task, PCA is however not able to use this information. The problem is more interesting if only part of the input data are labeled, i.e., in a semi-supervised setting. In this paper we propose a supervised PCA model called SPPCA and a semi-supervised PCA model called S2PPCA, both of which are extensions of a probabilistic PCA model. The proposed models are able to incorporate the label information into the projection phase, and can naturally handle multiple outputs (i.e., in multi-task learning problems). We derive an efficient EM learning algorithm for both models, and also provide theoretical justifications of the model behaviors. SPPCA and S2PPCA are compared with other supervised projection methods on various learning tasks, and show not only promising performance but also good scalability.

Averaging, maximum penalized likelihood and Bayesian estimation for improving Gaussian mixture probability density estimates

We apply the idea of averaging ensembles of estimators to probability density estimation. In particular, we use Gaussian mixture models which are important components in many neural-network applications. We investigate the performance of averaging using three data sets. For comparison, we employ two traditional regularization approaches, i.e., a maximum penalized likelihood approach and a Bayesian approach. In the maximum penalized likelihood approach we use penalty functions derived from conjugate Bayesian priors such that an expectation maximization (EM) algorithm can be used for training. In all experiments, the maximum penalized likelihood approach and averaging improved performance considerably if compared to a maximum likelihood approach. In two of the experiments, the maximum penalized likelihood approach outperformed averaging. In one experiment averaging was clearly superior. Our conclusion is that maximum penalized likelihood gives good results if the penalty term in the cost function is appropriate for the particular problem. If this is not the case, averaging is superior since it shows greater robustness by not relying on any particular prior assumption. The Bayesian approach worked very well on a low-dimensional toy problem but failed to give good performance in higher dimensional problems.

Mining the Semantic Web

In the Semantic Web vision of the World Wide Web, content will not only be accessible to humans but will also be available in machine interpretable form as ontological knowledge bases. Ontological knowledge bases enable formal querying and reasoning and, consequently, a main research focus has been the investigation of how deductive reasoning can be utilized in ontological representations to enable more advanced applications. However, purely logic methods have not yet proven to be very effective for several reasons: First, there still is the unsolved problem of scalability of reasoning to Web scale. Second, logical reasoning has problems with uncertain information, which is abundant on Semantic Web data due to its distributed and heterogeneous nature. Third, the construction of ontological knowledge bases suitable for advanced reasoning techniques is complex, which ultimately results in a lack of such expressive real-world data sets with large amounts of instance data. From another perspective, the more expressive structured representations open up new opportunities for data mining, knowledge extraction and machine learning techniques. If moving towards the idea that part of the knowledge already lies in the data, inductive methods appear promising, in particular since inductive methods can inherently handle noisy, inconsistent, uncertain and missing data. While there has been broad coverage of inducing concept structures from less structured sources (text, Web pages), like in ontology learning, given the problems mentioned above, we focus on new methods for dealing with Semantic Web knowledge bases, relying on statistical inference on their standard representations. We argue that machine learning research has to offer a wide variety of methods applicable to different expressivity levels of Semantic Web knowledge bases: ranging from weakly expressive but widely available knowledge bases in RDF to highly expressive first-order knowledge bases, this paper surveys statistical approaches to mining the Semantic Web. We specifically cover similarity and distance-based methods, kernel machines, multivariate prediction models, relational graphical models and first-order probabilistic learning approaches and discuss their applicability to Semantic Web representations. Finally we present selected experiments which were conducted on Semantic Web mining tasks for some of the algorithms presented before. This is intended to show the breadth and general potential of this exiting new research and application area for data mining.

Neural-network models for the blood glucose metabolism of a diabetic

We study the application of neural networks to modeling the blood glucose metabolism of a diabetic. In particular we consider recurrent neural networks and time series convolution neural networks which we compare to linear models and to nonlinear compartment models. We include a linear error model to take into account the uncertainty in the system and for handling missing blood glucose observations. Our results indicate that best performance can be achieved by the combination of the recurrent neural network and the linear error model.