Open Access

Skulls of 1,901 harbor seals from the North Sea were systematically investigated for dental, periodontal and cranial disorders as well as pathological changes of the temporomandibular joint (TMJ). Volumetric bone mineral density (vBMD) and microarchitecture of the mandible of a subsample of these specimens were analyzed in respect to age-related changes. Age at death of examined seals ranged from 1 week to 25 years. Most of the specimens were collected in 1988, when the population suffered from a phocine distemper virus epizootic. Therefore, it is assumed that, contrary to other museum collections, only little overrepresentation of pathological skeletal condition is present in the analyzed death sample. Age- and sex-related differences in the frequency and severity of pathological changes were observed in the dentition and the TMJ. Intravital tooth loss, tooth fracture and periapical lesions were recorded more frequently in male seals than in femals. Lesions consistent with temporomandibular joint osteoarthritis (TMJ-OA) also occurred more frequently in males, while lesion severity tended to be higher in female specimens. Severity of TMJ-OA lesions was positively correlated with age. Significant age-related changes in vBMD and several microarchitectural parameters were observed between individuals of the age classes “young juveniles” (0.5–10 months),“yearlings” (12–23 months),and “adults” (12–25 years),indicating an overall increase in cortical and trabecular area, cortical thickness as well as vBMD with age. For juvenile animals (≤ 23 months), positive correlations with age were observed for cortical area and thickness, trabecular separation, as well as vBMD. Negative correlations with age existed for trabecular number and thickness as well as for trabecular bone volume fraction (BV/TV) in the juveniles. The findings suggest a reduction in BV/TV with age,due to the bone trabeculae becoming thinner,less numerous and more widely spaced. This detailed knowledge of age-related changes in the structure and mineralization of bones is an important prerequisite for interpreting osseous changes in wild mammals caused by external factors, as such as exposure to environmental contaminants. The examination of large skeletal collections enables the observation also of rare pathological conditions. In the present investigation a case of anodontia,diagnosed as a manifestation of hypohidrotic ectodermal dysplasia was for the first time described in the harbor seal. In five juvenile Baltic grey seals,severe osteomyelitis of the jaws was described for the first time. The condition was attributed to disturbed dentin formation,presumably of genetic causation, in the affected individuals. The present study highlights the fact that systematic analyses of museum collections can provide important insights into the dental and skeletal pathology of wild mammals. These data can be used for reconstructing the health situation and living condition of past animal population.

In diesem Band enthalten: S. 1 – 34 Eika Ehme & Sabine Panzer-Krause Image und Stadtteilentwicklung: Attraktivierung innenstadtnaher Wohnviertel für Studierende und identitätsstiftende Maßnahmen am Beispiel der Hildesheimer Neustadt S. 35 – 65 Michelle Kieselstein Niedersächsische Lehrpfade – wie können traditionelle Bildungsinstrumente eine Bildung für nachhaltige Entwicklung ermöglichen? S. 66 – 90 Mischa Wittmar & Martin Sauerwein Geoökologische Untersuchungen zur Immissionsbelastung des Stadtwaldes Eilenriede (Hannover) S. 91 – 123 Moritz Sandner, Robin Stadtmann & Martin Sauerwein Möglichkeiten und Grenzen offener Fernerkundungsdaten und Open-Source- Software zur Landbedeckungsklassifikation des Nationalparks Cinque Terre (Italien) S. 124 – 129 Informationen aus dem Institut 2017 – 2018

The study of vegetation-plot data on a broad geographical scale is of increasing importance in vegetation science. It significantly contributes to the transnational characterisation of vegetation types as well as the better understanding of their large-scale patterns and to habitat typologies, which are important for decision-making processes in European nature conservation. I examined semi-natural, saline and brackish Baltic Sea grasslands which occur on sedimentary flats at the transition between land and sea. Their diverse vegetation is dependent on low intensity grazing (Dijkema 1990). This valuable part of the European cultural landscape (Küster 2004), which is recognized as Annex I priority habitat type (Natura 2000; European Commission 2013), underwent an overall decrease in quality and quantity within the last 150 years, which is frequently related to abandonment. Thus, the coastal grasslands of the Baltic Sea have been assessed as Endangered in the European Red List of Habitats (Janssen et al. 2016). Within this thesis I (i) developed a proposal to integrate vegetation data using non-standard scales into general vegetation analyses, (ii) characterised the vegetation of Baltic Sea grasslands on transnational level, (iii) regarded them from a North-west European perspective, (iv) discussed their nature conservation aspects on European scale, (v) investigated changes in their plant species composition and discussed its possible relation to cessation of grazing and (vi) formulated a monitoring concept important for management planning in nature conservation.

Machine learning is often confronted with the problem of learning prediction models on a set of observed data points. Given an expressive data set of the problem to solve, using powerful models and learning algorithms is only hindered by setting the right configurations for both. Unfortunately, the magnitude of the performance difference is large, which makes choosing right configurations an additional problem that is only solved by experienced practitioners. In this thesis, we will address the problem of hyperparameter optimization for machine learning and present ways to solve it. We firstly introduce the problem of supervised machine learning. We then discuss many examples of hyperparameter configurations that can be considered prior to learning the model. Afterwards, we introduce methods on finding the right configurations, especially those methods that work in the scheme of Bayesian optimization, which is a framework for optimizing black-box functions. Black-boxes are functions where for a given input one can only observe an output after running a costly procedure. Usually, in black-box optimization so-called surrogate models are learned to reconstruct the observations to then offer a prediction for unobserved configurations. Fortunately, recent outcomes show that transfering the knowledge across problems, for example by learning surrogates across different data sets being solved by the same model class, shows promising results. We tackle the problem of hyperparameter optimization in mainly two different ways. At first, we consider the problem of hyperparameter optimization as a recommendation problem, where we want to learn data set features as well as their interaction with the hyperparameter configurations as latent features in a factorization based approach. We build a surrogate model that is inspired by the complexity of neural networks as well as the ability to learn latent embeddings as in factorization machines. Secondly, as the amount of meta knowledge increases every day, surrogate models need to be scalable. We consider Gaussian processes, as they themselves are hyperparameter free and work very well in most hyperparameter optimization cases. Unfortunately, they are not scalable, as a matrix in the size of the number of data points has to be inverted for inference. We show various methods of simplifying a Gaussian process by using an ensemble of Gaussian process experts, which is much faster to learn due to its paralellization properties while still showing very competitive performance. We conclude the thesis by discussing the aspect of learning across problems in more detail than simply learning across different data sets. By learning hyperparameter performance across different models, we show that also model choice can be handled by the proposed algorithms. Additionally, we show that hyperparameter performance can even be transfered across different problem tasks, for example from classification to regression.

Viele Landoberflächen im Mediterranraum sind durch wellige bis steile Hänge charakterisiert. Nichtsdestotrotz erfuhren die hängigen Landschaftselemente seitens der geomorphologisch-bodengeographischen Forschung in der Vergangenheit nur wenig Beachtung. Der bodengeographische Kenntnisstand beruht weitgehend auf der Untersuchung von (fast-)ebenen Landschaften, wie Flussund Meeresterrassen oder Plateau- und Beckenlagen. Entsprechend nehmen bodengenetische Entwicklungskonzepte kaum Bezug zur Substratgenese durch Umlagerungsprozesse am Hang. Ziel der Arbeit ist daher, die Bedeutung der Substratbildung als Voraussetzung für die natürliche Bodenentwicklung auf Kalkgesteinen zu erfassen. Weiterführend wird anhand der Verbreitung von Substraten und der Abhängigkeit zur Pedogenese eine substratorientierte bodengenetische Modellvorstellung entwickelt. Im Rahmen der vorliegenden Arbeit werden geomorphologisch-bodengeographische Untersuchungen am Fallbeispiel eines Hangsystems auf Kalkgesteinen in der portugiesischen Estremadura durchgeführt. Basierend auf bodengeographischen Geländeuntersuchungen mit vertiefter Analyse des petrographischen Spektrums aus Kalkgesteinsvarietäten werden Stoffflussbahnen entlang der den Hang gliedernden Dellensysteme identifiziert, die die Verbreitung von allochthonen Substraten in Zwischenspeicherpositionen determinieren. Bestätigt durch einfache physikochemische Laboruntersuchungen kann eine eindeutige Abhängigkeit des vorzufindenden Bodenmosaiks zur Substratgenese im Meso- bzw. Mikrorelief gezeigt werden, die vorrangig durch spülaquatische Hangumlagerungsprozesse und untergeordnet auch durch gravitativen Versatz gesteuert ist. Gesamtmetallquotienten, granulometrische Analysen und exemplarisch quantitative Phytolithbestimmungen bestätigen die Substratgliederung der Pedone. Auch mit Hilfe einer adaptierten Faziesneutralen Lagenbeschreibung (aFNL) als feldmethodisches Werkzeug zur Schichtabgrenzung in mediterranen Kalksteinböden kann der Nachweis einer eindeutigen Koinzidenz der Bodenentwicklung an die Substratgenese auf Kalkgesteinen geführt werden. Die umfangreichen Aufschlussuntersuchungen zeigen auf der fast ebenen Hochfläche einen kleinräumigen Wechsel zwischen voll entwickelten Terra fuscae in allochthonen Substraten fuscae (ADHOC-ARBEITSGRUPPE BODEN 2005; IUSS WORKING GROUP WRB 2014: Calcaric Chromic Cambisol) und geringmächtigen (Locker-)Syrosemen oder (Para-) Rendzinen auf autochthonem Kalkgestein (AD-HOC-ARBEITSGRUPPE BODEN 2005; IUSS WORKING GROUP WRB 2014: Calcaric Regosols, (Renzic) Calcaric Leptosol). Die Verbreitung der Pedone entspricht der Verteilung von Karsttaschen und -schlotten gegenüber den Festgesteinsdurchragungen. Am Hang auf Kalkgesteinen sind ebenfalls allochthone Substrate (Hangsedimente) die Grundlage für die Verbreitung von Terra fuscae, die aber an die Verläufe von flachen Dellensystemen gebunden sind. Außerhalb der Dellen können auf autochthonen Carbonatgesteinen bzw. -aschen ebenfalls lediglich (Locker-)Syroseme oder (Para-)Rendzinen vorgefunden werden. In Unterhangbereichen treten meist eher flachgründige, selten auch mehrgliedrige, kolluviale Überdeckungen der Böden hinzu, die als Ergebnis anthropogen initiierter Bodenerosion der jüngeren Landschaftsgeschichte interpretiert werden. Zur Erklärung des anzutreffenden Bodenmosaiks wird im Rahmen der Arbeit ein substratgenetisch orientiertes, konsequent allochthonistisches Bodenentwicklungsmodell in Anlehnung an LORZ (2008a, 2008b) entworfen, das prinzipiell auch auf andere Hanglagen im Mediterranraum übertragbar ist. Aus der Befundlage ist zu konstatieren, dass das bodengeographische Muster regelhaft, ubiquitär und systematisch von der jeweiligen horizontalen und vertikalen Konfiguration der Ausgangssubstrate determiniert wird. Dabei wird Fern- und Lokalstaubeinträgen, die häufig als allochthone Komponente in bekannten autochthonen Pedogenesemodellen für den Mediterranraum einbezogen sind, nur eine untergeordnete Bedeutung beigemessen. Fortgeschrittenere Bodenentwicklungen, wie Terra fuscae, sind an allochthone Substrate (Hangsedimente; Regolithisierung) gebunden. Die Regolithisierung unter Einbeziehung von Pedosedimenten ist Voraussetzung für mächtigere und weiter entwickelte Böden (Terra fuscae) sowohl am Hang als auch auf der Hochfläche.

Infolge des gesellschaftlichen Wandels haben sich die Aufgaben und auch ihre Prioritätensetzung in der Pflegekinderhilfe verändert. Bspw. stellen sich die Ansprüche an erzieherische Fähigkeiten über die Pflegepersonen verfügen sollen, umfangreicher dar als noch vor Jahrzehnten. Zudem war eine Zusammenarbeit mit den Geburtseltern bis weit in das 20. Jahrhundert nicht vorgesehen. Gleichzeitig sind die bestehenden Beschreibungen der Pflegefamilie bis heute unscharf und in ständiger Veränderung darüber, wie eine (Pflege-) Familie aussehen soll, was sie charakterisiert, welche Aufgaben, welche Rollen vergeben werden. Die Forschung hat bisher zu wenig zur Weiterentwicklung beigetragen. Angesichts dessen bestehen Widersprüche für derartige Ausprägungen, wenn die sich wandelnden und als Norm festgelegten Lebensweisen nicht mit den sie umgebenden Umwelten harmonieren. Ihre Auswirkungen treten verstärkt auf der operativen Ebene hervor. Das vorliegende Forschungsprojekt hat Erstgespräche zwischen Fachkräften für Pflegekinder und Bewerbenden um ein Pflegekind evaluiert. Die in der Untersuchung freigelegten Ambivalenzen zeigen auf, dass sich einerseits die Anforderungen an den Pflegeauftrag gewandelt haben und andererseits diese veränderten Ansprüche offenbar in der Praxis noch nicht zufriedenstellend gelöst wurden. Deren Auftreten hat aber Auswirkungen u.a. auf die Herstellung eines gemeinsamen Arbeitsbündnisses in einem ersten Gespräch zur Aufnahme eines Pflegekindes sowie anschließend auf das Alltagsgeschehen während eines Pflegearrangements. Die Ergebnisse dieses Forschungsprojekts können als Grundlage dienen, ein transparentes und an den Bedarfen von Pflegefamilien orientiertes Vorgehen zu generieren.

Social media has become an integral part of numerous individuals as well as organizations, with many services being used frequently by a majority of people. Along with its widespread use, the amount of information explodes when people use these services. This demands for efficient tools as well as methods to assist data management and retrieval. Annotating resources by keywords, known as the tagging task, is a solution to improve categorizability and findability of resources. However, tagging is a human, time-consuming task, which requires the user's focus to figure out many keywords in a short moment and manually enter them into the system. To encourage users to tag their resources more correctly and frequently, tag recommendation is adopted into the social tagging systems to suggest relevant keywords for resources. In this thesis, we will address the problem of personalized tag recommendation for images and present ways to solve this problem by combining the advantages of the user relation with the images' content. In order to suggest tags for unobserved images, their visual contents are used to replace the index-based information of the image entity in the tagging relations. Because the limitation of low-level features does not show the "content" of images, we propose to utilize a deep learning based approach to learn high-level visual features concurrently with the scoring-tag estimator. For the tag predictor, a latent factor model or a multi-layer perceptron is selected to compute scores of tags by which the top selected tags are sorted in descending order. As a further development upon our findings, we examine the inside and outside context of images to enhance the accuracy of estimators. Regarding the image-inside context, we are motivated by the fact that objects, such as cars or cats are influential on the user's selection criteria. Regarding the image-outside context, the image's surrounding text contributes to the clarity of the image's content for different users. We consider these contextual features as a supporting part which is combined with the mainly visual representation to enhance the tag recommendation performance. Finally, as an additional technique, transfer learning is also adapted to support the proposed models to overcome the limitations of too small training data and boost up their performance. This thesis demonstrates the usefulness and versatility of deep learning approaches for tag recommendation and highlights the importance of the learned image's content in predicting personalized tags. Directions for future work include semantic enhancements to context-based representation and extensions of the content-aware approaches to different recommendation scenarios.

Automating machine learning by providing techniques that autonomously find the best algorithm, hyperparameter configuration and preprocessing is helpful for both researchers and practitioners. Therefore, it is not surprising that automated machine learning has become a very interesting field of research. Bayesian optimization has proven to be a very successful tool for automated machine learning. In the first part of the thesis we present different approaches to improve Bayesian optimization by means of transfer learning. We present three different ways of considering meta-knowledge in Bayesian optimization, i.e. search space pruning, initialization and transfer surrogate models. Finally, we present a general framework for Bayesian optimization combined with meta-learning and conduct a comparison among existing work on two different meta-data sets. A conclusion is that in particular the meta-target driven approaches provide better results. Choosing algorithm configurations based on the improvement on the meta-knowledge combined with the expected improvement yields best results. The second part of this thesis is more application-oriented. Bayesian optimization is applied to large data sets and used as a tool to participate in machine learning challenges. We compare its autonomous performance and its performance in combination with a human expert. At two ECML-PKDD Discovery Challenges, we are able to show that automated machine learning outperforms human machine learning experts. Finally, we present an approach that automates the process of creating an ensemble of several layers, different algorithms and hyperparameter configurations. These kinds of ensembles are jokingly called Frankenstein ensembles and proved their benefit on versatile data sets in many machine learning challenges. We compare our approach Automatic Frankensteining with the current state of the art for automated machine learning on 80 different data sets and can show that it outperforms them on the majority using the same training time. Furthermore, we compare Automatic Frankensteining on a large-scale data set to more than 3,500 machine learning expert teams and are able to outperform more than 3,000 of them within 12 CPU hours.

In this thesis we design and test Learning Analytics algorithms for personalized tasks' sequencing that suggests the next task to a student according to his/her specific needs. Our solution is based on a sequencing policy derived from the Vygotsky's Zone of Proximal Development (ZPD), which denes those tasks that are neither too easy not too dicult for the student. The sequencer, called Vygotsky Policy Sequencer (VPS), can identify tasks in the ZPD thanks to the information it receives from performance prediction algorithms able to estimate the knowledge of the student. Under this context we describe hereafter the thesis contributions. (1) A feasibility evaluation of domain independent Matrix Factorization applied in ITS for Performance Prediction. (2) An adaption and the related evaluation of a domain independent update for online learning Matrix Factorization in ITS. (3) A novel Matrix Factorization update method based on Kalman Filters approach. Two different updating functions are used: (a) a simple one considering the task just seen, and (b) one able to derive the skills' deficiency of the student. (4) A new method for offline testing of machine learning controlled sequencers by modeling simulated environment composed by a simulated students and tasks with continuous knowledge and score representation and different diffculty levels. (5) The design of a minimal invasive API for the lightweight integration of machine learning components in larger systems to minimize the risk of integration and the cost of expertise transfer. Profiting from all these contributions, the VPS was integrated in a commercial system and evaluated with 100 children over a month. The VPS showed comparable learning gains and perceived experience results with those of the ITS sequencer. Finally, thanks to its better modeling abilities, the students finish faster the assigned tasks.

Time series represent the most widely spread type of data, occurring in a myriad of application domains, ranging from physiological sensors up to astronomical light intensities. The classification of time-series is one of the most prominent challenges, which utilizes a recorded set of expert-labeled time-series, in order to automatically predict the label of future series without the need of an expert.The patterns of time-series are often shifted in time, have different scales, contain arbitrarily repeating patterns and exhibit local distortions/noise. In other cases, the differences among classes are attributed to small local segments, rather than the global structure. For those reasons, values corresponding to a particular time-stamp have different semantics on different time-series. We call this phenomena as intra-class variations. The lion's share of this thesis is composed of presenting new methods that can accurately classify time-series instances, by handling variations. The answer towards resolving the bottlenecks of intra-class variations relies on not using the time-series values as direct features. Instead, the approach of this thesis is to extract a set of features that, on one hand, represent all the variations of the data and, on the other hand, can boost classification accuracy. In other words, this thesis proposes a list of methods that addresses diverse aspects of intra-class variations. The first proposed approach is to generate new training instances, by transforming the support vectors of an SVM. The second approach decomposes time-series through a segment-wise convolutional factorization. The strategy involves learning a set of patterns and weights, whose product can approximate each sub-sequence of the time series. However, the main contribution of the thesis is the third approach, called shapelet learning, which utilizes the training labels during the learning process, i.e. the process is supervised. Since the features are learned on the training labels, there is a higher tendency of performing strongly in terms of predicting the testing labels. In addition, we present a fast alternative method for shapelet discovery. Our strategy is to prune segment candidates using a two step approach. First of all, we prune candidates based on their similarity towards previously considered candidates. Secondly, non-similar (hence diverse) candidates are selected only if the features they produce improve the classification results. The last two chapters of the thesis describes two methods that extract features from datasets having special characteristics. More concretely, we propose a classification method suited for series having missing values, as well as a method that extract features from time series having repetitive patterns.