TY - JOUR A1 - Ceulemans, Ruben A1 - Guill, Christian A1 - Gaedke, Ursula T1 - Top predators govern multitrophic diversity effects in tritrophic food webs JF - Ecology : a publication of the Ecological Society of America N2 - It is well known that functional diversity strongly affects ecosystem functioning. However, even in rather simple model communities consisting of only two or, at best, three trophic levels, the relationship between multitrophic functional diversity and ecosystem functioning appears difficult to generalize, because of its high contextuality. In this study, we considered several differently structured tritrophic food webs, in which the amount of functional diversity was varied independently on each trophic level. To achieve generalizable results, largely independent of parametrization, we examined the outcomes of 128,000 parameter combinations sampled from ecologically plausible intervals, with each tested for 200 randomly sampled initial conditions. Analysis of our data was done by training a random forest model. This method enables the identification of complex patterns in the data through partial dependence graphs, and the comparison of the relative influence of model parameters, including the degree of diversity, on food-web properties. We found that bottom-up and top-down effects cascade simultaneously throughout the food web, intimately linking the effects of functional diversity of any trophic level to the amount of diversity of other trophic levels, which may explain the difficulty in unifying results from previous studies. Strikingly, only with high diversity throughout the whole food web, different interactions synergize to ensure efficient exploitation of the available nutrients and efficient biomass transfer to higher trophic levels, ultimately leading to a high biomass and production on the top level. The temporal variation of biomass showed a more complex pattern with increasing multitrophic diversity: while the system initially became less variable, eventually the temporal variation rose again because of the increasingly complex dynamical patterns. Importantly, top predator diversity and food-web parameters affecting the top trophic level were of highest importance to determine the biomass and temporal variability of any trophic level. Overall, our study reveals that the mechanisms by which diversity influences ecosystem functioning are affected by every part of the food web, hampering the extrapolation of insights from simple monotrophic or bitrophic systems to complex natural food webs. KW - food-web efficiency KW - functional diversity KW - machine learning KW - nutrient KW - exploitation KW - production KW - random forest KW - temporal variability KW - top KW - predator KW - trait diversity Y1 - 2021 U6 - https://doi.org/10.1002/ecy.3379 SN - 0012-9658 SN - 1939-9170 VL - 102 IS - 7 PB - Wiley CY - Hoboken ER - TY - JOUR A1 - Steinberg, Andreas A1 - Vasyura-Bathke, Hannes A1 - Gaebler, Peter Jost A1 - Ohrnberger, Matthias A1 - Ceranna, Lars T1 - Estimation of seismic moment tensors using variational inference machine learning JF - Journal of geophysical research : Solid earth N2 - We present an approach for rapidly estimating full moment tensors of earthquakes and their parameter uncertainties based on short time windows of recorded seismic waveform data by considering deep learning of Bayesian Neural Networks (BNNs). The individual neural networks are trained on synthetic seismic waveform data and corresponding known earthquake moment-tensor parameters. A monitoring volume has been predefined to form a three-dimensional grid of locations and to train a BNN for each grid point. Variational inference on several of these networks allows us to consider several sources of error and how they affect the estimated full moment-tensor parameters and their uncertainties. In particular, we demonstrate how estimated parameter distributions are affected by uncertainties in the earthquake centroid location in space and time as well as in the assumed Earth structure model. We apply our approach as a proof of concept on seismic waveform recordings of aftershocks of the Ridgecrest 2019 earthquake with moment magnitudes ranging from Mw 2.7 to Mw 5.5. Overall, good agreement has been achieved between inferred parameter ensembles and independently estimated parameters using classical methods. Our developed approach is fast and robust, and therefore, suitable for down-stream analyses that need rapid estimates of the source mechanism for a large number of earthquakes. KW - seismology KW - machine learning KW - earthquake source KW - moment tensor KW - full KW - waveform Y1 - 2021 U6 - https://doi.org/10.1029/2021JB022685 SN - 2169-9313 SN - 2169-9356 VL - 126 IS - 10 PB - American Geophysical Union CY - Washington ER - TY - JOUR A1 - Cope, Justin L. A1 - Baukmann, Hannes A. A1 - Klinger, Jörn E. A1 - Ravarani, Charles N. J. A1 - Böttinger, Erwin A1 - Konigorski, Stefan A1 - Schmidt, Marco F. T1 - Interaction-based feature selection algorithm outperforms polygenic risk score in predicting Parkinson’s Disease status JF - Frontiers in genetics N2 - Polygenic risk scores (PRS) aggregating results from genome-wide association studies are the state of the art in the prediction of susceptibility to complex traits or diseases, yet their predictive performance is limited for various reasons, not least of which is their failure to incorporate the effects of gene-gene interactions. Novel machine learning algorithms that use large amounts of data promise to find gene-gene interactions in order to build models with better predictive performance than PRS. Here, we present a data preprocessing step by using data-mining of contextual information to reduce the number of features, enabling machine learning algorithms to identify gene-gene interactions. We applied our approach to the Parkinson's Progression Markers Initiative (PPMI) dataset, an observational clinical study of 471 genotyped subjects (368 cases and 152 controls). With an AUC of 0.85 (95% CI = [0.72; 0.96]), the interaction-based prediction model outperforms the PRS (AUC of 0.58 (95% CI = [0.42; 0.81])). Furthermore, feature importance analysis of the model provided insights into the mechanism of Parkinson's disease. For instance, the model revealed an interaction of previously described drug target candidate genes TMEM175 and GAPDHP25. These results demonstrate that interaction-based machine learning models can improve genetic prediction models and might provide an answer to the missing heritability problem. KW - epistasis KW - machine learning KW - feature selection KW - parkinson's disease KW - PPMI (parkinson's progression markers initiative) Y1 - 2021 U6 - https://doi.org/10.3389/fgene.2021.744557 SN - 1664-8021 VL - 12 PB - Frontiers Media CY - Lausanne ER - TY - GEN A1 - Konak, Orhan A1 - Wegner, Pit A1 - Arnrich, Bert T1 - IMU-Based Movement Trajectory Heatmaps for Human Activity Recognition T2 - Postprints der Universität Potsdam : Reihe der Digital Engineering Fakultät N2 - Recent trends in ubiquitous computing have led to a proliferation of studies that focus on human activity recognition (HAR) utilizing inertial sensor data that consist of acceleration, orientation and angular velocity. However, the performances of such approaches are limited by the amount of annotated training data, especially in fields where annotating data is highly time-consuming and requires specialized professionals, such as in healthcare. In image classification, this limitation has been mitigated by powerful oversampling techniques such as data augmentation. Using this technique, this work evaluates to what extent transforming inertial sensor data into movement trajectories and into 2D heatmap images can be advantageous for HAR when data are scarce. A convolutional long short-term memory (ConvLSTM) network that incorporates spatiotemporal correlations was used to classify the heatmap images. Evaluation was carried out on Deep Inertial Poser (DIP), a known dataset composed of inertial sensor data. The results obtained suggest that for datasets with large numbers of subjects, using state-of-the-art methods remains the best alternative. However, a performance advantage was achieved for small datasets, which is usually the case in healthcare. Moreover, movement trajectories provide a visual representation of human activities, which can help researchers to better interpret and analyze motion patterns. T3 - Zweitveröffentlichungen der Universität Potsdam : Reihe der Digital Engineering Fakultät - 4 KW - human activity recognition KW - image processing KW - machine learning KW - sensor data Y1 - 2021 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-487799 IS - 4 ER - TY - THES A1 - Lazaridou, Konstantina T1 - Revealing hidden patterns in political news and social media with machine learning T1 - Aufdecken versteckter Muster in politischen Nachrichten und sozialen Medien mit Hilfe von maschinellem Lernen N2 - As part of our everyday life we consume breaking news and interpret it based on our own viewpoints and beliefs. We have easy access to online social networking platforms and news media websites, where we inform ourselves about current affairs and often post about our own views, such as in news comments or social media posts. The media ecosystem enables opinions and facts to travel from news sources to news readers, from news article commenters to other readers, from social network users to their followers, etc. The views of the world many of us have depend on the information we receive via online news and social media. Hence, it is essential to maintain accurate, reliable and objective online content to ensure democracy and verity on the Web. To this end, we contribute to a trustworthy media ecosystem by analyzing news and social media in the context of politics to ensure that media serves the public interest. In this thesis, we use text mining, natural language processing and machine learning techniques to reveal underlying patterns in political news articles and political discourse in social networks. Mainstream news sources typically cover a great amount of the same news stories every day, but they often place them in a different context or report them from different perspectives. In this thesis, we are interested in how distinct and predictable newspaper journalists are, in the way they report the news, as a means to understand and identify their different political beliefs. To this end, we propose two models that classify text from news articles to their respective original news source, i.e., reported speech and also news comments. Our goal is to capture systematic quoting and commenting patterns by journalists and news commenters respectively, which can lead us to the newspaper where the quotes and comments are originally published. Predicting news sources can help us understand the potential subjective nature behind news storytelling and the magnitude of this phenomenon. Revealing this hidden knowledge can restore our trust in media by advancing transparency and diversity in the news. Media bias can be expressed in various subtle ways in the text and it is often challenging to identify these bias manifestations correctly, even for humans. However, media experts, e.g., journalists, are a powerful resource that can help us overcome the vague definition of political media bias and they can also assist automatic learners to find the hidden bias in the text. Due to the enormous technological advances in artificial intelligence, we hypothesize that identifying political bias in the news could be achieved through the combination of sophisticated deep learning modelsxi and domain expertise. Therefore, our second contribution is a high-quality and reliable news dataset annotated by journalists for political bias and a state-of-the-art solution for this task based on curriculum learning. Our aim is to discover whether domain expertise is necessary for this task and to provide an automatic solution for this traditionally manually-solved problem. User generated content is fundamentally different from news articles, e.g., messages are shorter, they are often personal and opinionated, they refer to specific topics and persons, etc. Regarding political and socio-economic news, individuals in online communities make use of social networks to keep their peers up-to-date and to share their own views on ongoing affairs. We believe that social media is also an as powerful instrument for information flow as the news sources are, and we use its unique characteristic of rapid news coverage for two applications. We analyze Twitter messages and debate transcripts during live political presidential debates to automatically predict the topics that Twitter users discuss. Our goal is to discover the favoured topics in online communities on the dates of political events as a way to understand the political subjects of public interest. With the up-to-dateness of microblogs, an additional opportunity emerges, namely to use social media posts and leverage the real-time verity about discussed individuals to find their locations. That is, given a person of interest that is mentioned in online discussions, we use the wisdom of the crowd to automatically track her physical locations over time. We evaluate our approach in the context of politics, i.e., we predict the locations of US politicians as a proof of concept for important use cases, such as to track people that are national risks, e.g., warlords and wanted criminals. N2 - Als festen Bestandteil unseres täglichen Lebens konsumieren wir aktuelle Nachrichten und interpretieren sie basierend auf unseren eigenen Ansichten und Überzeugungen. Wir haben einfachen Zugang zu sozialen Netzwerken und Online-Nachrichtenportalen, auf denen wir uns über aktuelle Angelegenheiten informieren und eigene Ansichten teilen, wie zum Beispiel mit Nachrichtenkommentaren oder Social-Media-Posts. Das Medien-Ökosystem ermöglicht es zum Beispiel, dass Meinungen und Fakten von Nachrichtenquellen zu Lesern, von Kommentatoren zu anderen Lesern oder von Nutzern sozialer Netzwerke zu ihren Anhängern gelangen. Die Weltsicht hängt für viele von uns von Informationen ab, die wir über Online-Nachrichten und soziale Medien erhalten. Hierfür ist es wichtig genaue, zuverlässige und objektive Inhalte zuzusichern, um die Demokratie und Wahrheit im Web gewährleisten zu können. Um zu einem vertrauenswürdigen Medien-Ökosystem beizutragen, analysieren wir Nachrichten und soziale Medien im politischen Kontext und stellen sicher, dass die Medien dem öffentlichen Interesse dienen. In dieser Arbeit verwenden wir Techniken der Computerlinguistik, des maschinellen Lernens und des Text Minings, um zugrunde liegende Muster in politischen Nachrichtenartikel und im politischen Diskurs in sozialen Netzwerken aufzudecken. Mainstream-Nachrichtenquellen decken täglich üb­li­cher­wei­se eine große Anzahl derselben Nachrichten ab, aber sie stellen diese oft in einem anderen Kontext dar oder berichten aus unterschiedlichen Sichtweisen. In dieser Arbeit wird untersucht, wie individuell und vorhersehbar Zeitungsjournalisten in der Art der Berichterstattung sind, um die unterschiedlichen politischen Überzeugungen zu identifizieren und zu verstehen. Zu diesem Zweck schlagen wir zwei Modelle vor, die Text aus Nachrichtenartikeln klassifizieren und ihrer jeweiligen ursprünglichen Nachrichtenquelle zuordnen, insbesondere basierend auf Zitaten und Nachrichtenkommentaren. Unser Ziel ist es, systematische Zitierungs- und Kommentierungsmuster von Journalisten bzw. Nachrichtenkommentatoren zu erfassen, was uns zu der Zeitung führen kann, in der die Zitate und Kommentare ursprünglich veröffentlicht wurden. Die Vorhersage von Nachrichtenquellen kann uns helfen, die potenziell subjektive Natur hinter dem “Storytelling” und dem Ausmaß dieses Phänomens zu verstehen. Das enthüllen jenes verborgenen Wissens kann unser Vertrauen in die Medien wiederherstellen, indem es Transparenz und Vielfalt in den Nachrichten fördert. Politische Tendenzen in der Medienberichterstattung können textuell auf verschiedene subtile Arten ausgedrückt werden und es ist selbst für Menschen oft schwierig deren Manifestierung korrekt zu identifizieren. Medienexperten wie Journalisten, sind jedoch eine gute Ressource, die uns helfen kann, die vage Definition der politischen Medien Bias zu überwinden und sie können ebenfalls dabei helfen automatischen Modellen beizubringen, versteckten Bias im Text aufzudecken. Aufgrund der enormen technologischen Fortschritte im Bereich der künstlichen Intelligenz nehmen wir an, dass die Identifizierung politischer Vorurteile in den Nachrichten durch die Kombination aus ausgefeilten Deep-Learning-Modellen und Fachkenntnissen erreicht werden kann. Daher ist unser zweiter Beitrag ein qualitativ hochwertiger und zuverlässiger Nachrichtendatensatz, der von Journalisten in Bezug auf politischen Bias annotiert wurde und ein hochmoderner Algorithmus zur Lösung dieser Aufgabe, der auf dem Prinzip des “curriculum learning” basiert. Unser Ziel ist es herauszufinden, ob Domänenwissen für diese Aufgabe erforderlich ist und eine automatische Lösung für dieses traditionell manuell bearbeitete Problem bereitzustellen. Nutzergenerierte Inhalte unterscheiden sich grundlegend von Nachrichtenartikeln. Zum Beispiel sind Botschaften oft kürzer, persönlich und dogmatisch und sie beziehen sich oft auf spezifische Themen und Personen. In Bezug auf politische und sozioökonomische Nachrichten verwenden Individuen oft soziale Netzwerke, um andere Nutzer in ihrer In­te­r­es­sens­grup­pe auf dem Laufenden zu halten und ihre persönlichen Ansichten über aktuelle Angelegenheiten zu teilen. Wir glauben, dass soziale Medien auch ein gleichermaßen leistungsfähiges Instrument für den Informationsfluss sind wie Online-Zeitungen. Daher verwenden wir ihre einzigartige Eigenschaft der schnellen Berichterstattung für zwei Anwendungen. Wir analysieren Twitter-Nachrichten und Transkripte von politischen Live-Debatten zur Präsidentschaftswahl um Themen zu klassifizieren, die von der Nutzergemeinde diskutiert werden. Unser Ziel ist es die bevorzugten Themen zu identifizieren, die in Online-Gemeinschaften zu den Terminen politischer Ereignisse diskutiert werden um die Themen von öffentlichem Interesse zu verstehen. Durch die Aktualität von Microblogs ergibt sich die zusätzliche Möglichkeit Beiträge aus sozialen Medien zu nutzen um Echtzeit-Informationen über besprochene Personen zu finden und ihre physischen Positionen zu bestimmen. Das heißt, bei einer Person von öffentlichem Interesse, die in Online-Diskussionen erwähnt wird, verwenden wir die Schwarmintelligenz der Nutzerbasis, um ihren Standort im Verlauf der Zeit automatisch zu verfolgen. Wir untersuchen unseren Ansatz im politischen Kontext, indem wir die Standorte von US-Politikern während des Präsidentschaftswahlkampfes voraussagen. Mit diesem Ansatz bieten wir eine Machbarkeitsstudie für andere wichtige Anwendungsfälle, beispielsweise um Menschen zu verfolgen, die ein nationales Risiko darstellen, wie Kriegsherren und gesuchte Kriminelle. KW - media bias KW - news KW - politics KW - machine learning KW - maschinelles Lernen KW - Medien Bias KW - Nachrichten KW - Politik Y1 - 2021 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-502734 ER - TY - THES A1 - Loster, Michael T1 - Knowledge base construction with machine learning methods T1 - Aufbau von Wissensbasen mit Methoden des maschinellen Lernens N2 - Modern knowledge bases contain and organize knowledge from many different topic areas. Apart from specific entity information, they also store information about their relationships amongst each other. Combining this information results in a knowledge graph that can be particularly helpful in cases where relationships are of central importance. Among other applications, modern risk assessment in the financial sector can benefit from the inherent network structure of such knowledge graphs by assessing the consequences and risks of certain events, such as corporate insolvencies or fraudulent behavior, based on the underlying network structure. As public knowledge bases often do not contain the necessary information for the analysis of such scenarios, the need arises to create and maintain dedicated domain-specific knowledge bases. This thesis investigates the process of creating domain-specific knowledge bases from structured and unstructured data sources. In particular, it addresses the topics of named entity recognition (NER), duplicate detection, and knowledge validation, which represent essential steps in the construction of knowledge bases. As such, we present a novel method for duplicate detection based on a Siamese neural network that is able to learn a dataset-specific similarity measure which is used to identify duplicates. Using the specialized network architecture, we design and implement a knowledge transfer between two deduplication networks, which leads to significant performance improvements and a reduction of required training data. Furthermore, we propose a named entity recognition approach that is able to identify company names by integrating external knowledge in the form of dictionaries into the training process of a conditional random field classifier. In this context, we study the effects of different dictionaries on the performance of the NER classifier. We show that both the inclusion of domain knowledge as well as the generation and use of alias names results in significant performance improvements. For the validation of knowledge represented in a knowledge base, we introduce Colt, a framework for knowledge validation based on the interactive quality assessment of logical rules. In its most expressive implementation, we combine Gaussian processes with neural networks to create Colt-GP, an interactive algorithm for learning rule models. Unlike other approaches, Colt-GP uses knowledge graph embeddings and user feedback to cope with data quality issues of knowledge bases. The learned rule model can be used to conditionally apply a rule and assess its quality. Finally, we present CurEx, a prototypical system for building domain-specific knowledge bases from structured and unstructured data sources. Its modular design is based on scalable technologies, which, in addition to processing large datasets, ensures that the modules can be easily exchanged or extended. CurEx offers multiple user interfaces, each tailored to the individual needs of a specific user group and is fully compatible with the Colt framework, which can be used as part of the system. We conduct a wide range of experiments with different datasets to determine the strengths and weaknesses of the proposed methods. To ensure the validity of our results, we compare the proposed methods with competing approaches. N2 - Moderne Wissensbasen enthalten und organisieren das Wissen vieler unterschiedlicher Themengebiete. So speichern sie neben bestimmten Entitätsinformationen auch Informationen über deren Beziehungen untereinander. Kombiniert man diese Informationen, ergibt sich ein Wissensgraph, der besonders in Anwendungsfällen hilfreich sein kann, in denen Entitätsbeziehungen von zentraler Bedeutung sind. Neben anderen Anwendungen, kann die moderne Risikobewertung im Finanzsektor von der inhärenten Netzwerkstruktur solcher Wissensgraphen profitieren, indem Folgen und Risiken bestimmter Ereignisse, wie z.B. Unternehmensinsolvenzen oder betrügerisches Verhalten, auf Grundlage des zugrundeliegenden Netzwerks bewertet werden. Da öffentliche Wissensbasen oft nicht die notwendigen Informationen zur Analyse solcher Szenarien enthalten, entsteht die Notwendigkeit, spezielle domänenspezifische Wissensbasen zu erstellen und zu pflegen. Diese Arbeit untersucht den Erstellungsprozess von domänenspezifischen Wissensdatenbanken aus strukturierten und unstrukturierten Datenquellen. Im speziellen befasst sie sich mit den Bereichen Named Entity Recognition (NER), Duplikaterkennung sowie Wissensvalidierung, die wesentliche Prozessschritte beim Aufbau von Wissensbasen darstellen. Wir stellen eine neuartige Methode zur Duplikaterkennung vor, die auf Siamesischen Neuronalen Netzwerken basiert und in der Lage ist, ein datensatz-spezifisches Ähnlichkeitsmaß zu erlernen, welches wir zur Identifikation von Duplikaten verwenden. Unter Verwendung einer speziellen Netzwerkarchitektur entwerfen und setzen wir einen Wissenstransfer zwischen Deduplizierungsnetzwerken um, der zu erheblichen Leistungsverbesserungen und einer Reduktion der benötigten Trainingsdaten führt. Weiterhin schlagen wir einen Ansatz zur Erkennung benannter Entitäten (Named Entity Recognition (NER)) vor, der in der Lage ist, Firmennamen zu identifizieren, indem externes Wissen in Form von Wörterbüchern in den Trainingsprozess eines Conditional Random Field Klassifizierers integriert wird. In diesem Zusammenhang untersuchen wir die Auswirkungen verschiedener Wörterbücher auf die Leistungsfähigkeit des NER-Klassifikators und zeigen, dass sowohl die Einbeziehung von Domänenwissen als auch die Generierung und Verwendung von Alias-Namen zu einer signifikanten Leistungssteigerung führt. Zur Validierung der in einer Wissensbasis enthaltenen Fakten stellen wir mit COLT ein Framework zur Wissensvalidierung vor, dass auf der interaktiven Qualitätsbewertung von logischen Regeln basiert. In seiner ausdrucksstärksten Implementierung kombinieren wir Gauß'sche Prozesse mit neuronalen Netzen, um so COLT-GP, einen interaktiven Algorithmus zum Erlernen von Regelmodellen, zu erzeugen. Im Gegensatz zu anderen Ansätzen verwendet COLT-GP Knowledge Graph Embeddings und Nutzer-Feedback, um Datenqualitätsprobleme des zugrunde liegenden Wissensgraphen zu behandeln. Das von COLT-GP erlernte Regelmodell kann sowohl zur bedingten Anwendung einer Regel als auch zur Bewertung ihrer Qualität verwendet werden. Schließlich stellen wir mit CurEx, ein prototypisches System zum Aufbau domänenspezifischer Wissensbasen aus strukturierten und unstrukturierten Datenquellen, vor. Sein modularer Aufbau basiert auf skalierbaren Technologien, die neben der Verarbeitung großer Datenmengen auch die einfache Austausch- und Erweiterbarkeit einzelner Module gewährleisten. CurEx bietet mehrere Benutzeroberflächen, die jeweils auf die individuellen Bedürfnisse bestimmter Benutzergruppen zugeschnitten sind. Darüber hinaus ist es vollständig kompatibel zum COLT-Framework, was als Teil des Systems verwendet werden kann. Wir führen eine Vielzahl von Experimenten mit unterschiedlichen Datensätzen durch, um die Stärken und Schwächen der vorgeschlagenen Methoden zu ermitteln. Zudem vergleichen wir die vorgeschlagenen Methoden mit konkurrierenden Ansätzen, um die Validität unserer Ergebnisse sicherzustellen. KW - machine learning KW - deep kernel learning KW - knowledge base construction KW - knowledge base KW - knowledge graph KW - deduplication KW - siamese neural networks KW - duplicate detection KW - entity resolution KW - transfer learning KW - knowledge transfer KW - entity linking KW - knowledge validation KW - logic rules KW - named entity recognition KW - curex KW - Curex KW - Deduplikation KW - Deep Kernel Learning KW - Duplikaterkennung KW - Entitätsverknüpfung KW - Entitätsauflösung KW - Wissensbasis KW - Konstruktion von Wissensbasen KW - Wissensgraph KW - Wissenstransfer KW - Wissensvalidierung KW - logische Regeln KW - maschinelles Lernen KW - named entity recognition KW - Siamesische Neuronale Netzwerke KW - Transferlernen Y1 - 2021 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-501459 ER - TY - JOUR A1 - Ayzel, Georgy T1 - Deep neural networks in hydrology BT - the new generation of universal and efficient models BT - новое поколение универсальных и эффективных моделей JF - Vestnik of Saint Petersburg University. Earth Sciences N2 - For around a decade, deep learning - the sub-field of machine learning that refers to artificial neural networks comprised of many computational layers - modifies the landscape of statistical model development in many research areas, such as image classification, machine translation, and speech recognition. Geoscientific disciplines in general and the field of hydrology in particular, also do not stand aside from this movement. Recently, the proliferation of modern deep learning-based techniques and methods has been actively gaining popularity for solving a wide range of hydrological problems: modeling and forecasting of river runoff, hydrological model parameters regionalization, assessment of available water resources. identification of the main drivers of the recent change in water balance components. This growing popularity of deep neural networks is primarily due to their high universality and efficiency. The presented qualities, together with the rapidly growing amount of accumulated environmental information, as well as increasing availability of computing facilities and resources, allow us to speak about deep neural networks as a new generation of mathematical models designed to, if not to replace existing solutions, but significantly enrich the field of geophysical processes modeling. This paper provides a brief overview of the current state of the field of development and application of deep neural networks in hydrology. Also in the following study, the qualitative long-term forecast regarding the development of deep learning technology for managing the corresponding hydrological modeling challenges is provided based on the use of "Gartner Hype Curve", which in the general details describes a life cycle of modern technologies. N2 - В течение последнего десятилетия глубокое обучение - область машинного обучения, относящаяся к искусственным нейронным сетям, состоящим из множества вычислительных слоев, - изменяет ландшафт развития статистических моделей во многих областях исследований, таких как классификация изображений, машинный перевод, распознавание речи. Географические науки, а также входящая в их состав область исследования гидрологии суши, не стоят в стороне от этого движения. В последнее время применение современных технологий и методов глубокого обучения активно набирает популярность для решения широкого спектра гидрологических задач: моделирования и прогнозирования речного стока, районирования модельных параметров, оценки располагаемых водных ресурсов, идентификации факторов, влияющих на современные изменения водного режима. Такой рост популярности глубоких нейронных сетей продиктован прежде всего их высокой универсальностью и эффективностью. Представленные качества в совокупности с быстрорастущим количеством накопленной информации о состоянии окружающей среды, а также ростом доступности вычислительных средств и ресурсов, позволяют говорить о глубоких нейронных сетях как о новом поколении математических моделей, призванных если не заменить существующие решения, то значительно обогатить область моделирования геофизических процессов. В данной работе представлен краткий обзор текущего состояния области разработки и применения глубоких нейронных сетей в гидрологии. Также в работе предложен качественный долгосрочный прогноз развития технологии глубокого обучения для решения задач гидрологического моделирования на основе использования «кривой ажиотажа Гартнера», в общих чертах описывающей жизненный цикл современных технологий. T2 - Глубокие нейронные сети в гидрологии KW - deep neural networks KW - deep learning KW - machine learning KW - hydrology KW - modeling KW - глубокие нейронные сети KW - глубокое обучение KW - машинное обучение KW - гидрология KW - моделирование Y1 - 2021 U6 - https://doi.org/10.21638/spbu07.2021.101 SN - 2541-9668 SN - 2587-585X VL - 66 IS - 1 SP - 5 EP - 18 PB - Univ. Press CY - St. Petersburg ER - TY - JOUR A1 - Brandes, Stefanie A1 - Sicks, Florian A1 - Berger, Anne T1 - Behaviour classification on giraffes (Giraffa camelopardalis) using machine learning algorithms on triaxial acceleration data of two commonly used GPS devices and its possible application for their management and conservation JF - Sensors N2 - Averting today's loss of biodiversity and ecosystem services can be achieved through conservation efforts, especially of keystone species. Giraffes (Giraffa camelopardalis) play an important role in sustaining Africa's ecosystems, but are 'vulnerable' according to the IUCN Red List since 2016. Monitoring an animal's behavior in the wild helps to develop and assess their conservation management. One mechanism for remote tracking of wildlife behavior is to attach accelerometers to animals to record their body movement. We tested two different commercially available high-resolution accelerometers, e-obs and Africa Wildlife Tracking (AWT), attached to the top of the heads of three captive giraffes and analyzed the accuracy of automatic behavior classifications, focused on the Random Forests algorithm. For both accelerometers, behaviors of lower variety in head and neck movements could be better predicted (i.e., feeding above eye level, mean prediction accuracy e-obs/AWT: 97.6%/99.7%; drinking: 96.7%/97.0%) than those with a higher variety of body postures (such as standing: 90.7-91.0%/75.2-76.7%; rumination: 89.6-91.6%/53.5-86.5%). Nonetheless both devices come with limitations and especially the AWT needs technological adaptations before applying it on animals in the wild. Nevertheless, looking at the prediction results, both are promising accelerometers for behavioral classification of giraffes. Therefore, these devices when applied to free-ranging animals, in combination with GPS tracking, can contribute greatly to the conservation of giraffes. KW - giraffe KW - triaxial acceleration KW - machine learning KW - random forests KW - behavior classification KW - giraffe conservation Y1 - 2021 U6 - https://doi.org/10.3390/s21062229 SN - 1424-8220 VL - 21 IS - 6 PB - MDPI CY - Basel ER - TY - JOUR A1 - Hampf, Anna A1 - Nendel, Claas A1 - Strey, Simone A1 - Strey, Robert T1 - Biotic yield losses in the Southern Amazon, Brazil BT - making use of smartphone-assisted plant disease diagnosis data JF - Frontiers in plant science : FPLS N2 - Pathogens and animal pests (P&A) are a major threat to global food security as they directly affect the quantity and quality of food. The Southern Amazon, Brazil's largest domestic region for soybean, maize and cotton production, is particularly vulnerable to the outbreak of P&A due to its (sub)tropical climate and intensive farming systems. However, little is known about the spatial distribution of P&A and the related yield losses. Machine learning approaches for the automated recognition of plant diseases can help to overcome this research gap. The main objectives of this study are to (1) evaluate the performance of Convolutional Neural Networks (ConvNets) in classifying P&A, (2) map the spatial distribution of P&A in the Southern Amazon, and (3) quantify perceived yield and economic losses for the main soybean and maize P&A. The objectives were addressed by making use of data collected with the smartphone application Plantix. The core of the app's functioning is the automated recognition of plant diseases via ConvNets. Data on expected yield losses were gathered through a short survey included in an "expert" version of the application, which was distributed among agronomists. Between 2016 and 2020, Plantix users collected approximately 78,000 georeferenced P&A images in the Southern Amazon. The study results indicate a high performance of the trained ConvNets in classifying 420 different crop-disease combinations. Spatial distribution maps and expert-based yield loss estimates indicate that maize rust, bacterial stalk rot and the fall armyworm are among the most severe maize P&A, whereas soybean is mainly affected by P&A like anthracnose, downy mildew, frogeye leaf spot, stink bugs and brown spot. Perceived soybean and maize yield losses amount to 12 and 16%, respectively, resulting in annual yield losses of approximately 3.75 million tonnes for each crop and economic losses of US$2 billion for both crops together. The high level of accuracy of the trained ConvNets, when paired with widespread use from following a citizen-science approach, results in a data source that will shed new light on yield loss estimates, e.g., for the analysis of yield gaps and the development of measures to minimise them. KW - plant pathology KW - animal pests KW - pathogens KW - machine learning KW - digital KW - image processing KW - disease diagnosis KW - crowdsourcing KW - crop losses Y1 - 2021 U6 - https://doi.org/10.3389/fpls.2021.621168 SN - 1664-462X VL - 12 PB - Frontiers Media CY - Lausanne ER - TY - JOUR A1 - Vaid, Akhil A1 - Chan, Lili A1 - Chaudhary, Kumardeep A1 - Jaladanki, Suraj K. A1 - Paranjpe, Ishan A1 - Russak, Adam J. A1 - Kia, Arash A1 - Timsina, Prem A1 - Levin, Matthew A. A1 - He, John Cijiang A1 - Böttinger, Erwin A1 - Charney, Alexander W. A1 - Fayad, Zahi A. A1 - Coca, Steven G. A1 - Glicksberg, Benjamin S. A1 - Nadkarni, Girish N. T1 - Predictive approaches for acute dialysis requirement and death in COVID-19 JF - Clinical journal of the American Society of Nephrology : CJASN N2 - Background and objectives AKI treated with dialysis initiation is a common complication of coronavirus disease 2019 (COVID-19) among hospitalized patients. However, dialysis supplies and personnel are often limited. Design, setting, participants, & measurements Using data from adult patients hospitalized with COVID-19 from five hospitals from theMount Sinai Health System who were admitted between March 10 and December 26, 2020, we developed and validated several models (logistic regression, Least Absolute Shrinkage and Selection Operator (LASSO), random forest, and eXtreme GradientBoosting [XGBoost; with and without imputation]) for predicting treatment with dialysis or death at various time horizons (1, 3, 5, and 7 days) after hospital admission. Patients admitted to theMount Sinai Hospital were used for internal validation, whereas the other hospitals formed part of the external validation cohort. Features included demographics, comorbidities, and laboratory and vital signs within 12 hours of hospital admission. Results A total of 6093 patients (2442 in training and 3651 in external validation) were included in the final cohort. Of the different modeling approaches used, XGBoost without imputation had the highest area under the receiver operating characteristic (AUROC) curve on internal validation (range of 0.93-0.98) and area under the precisionrecall curve (AUPRC; range of 0.78-0.82) for all time points. XGBoost without imputation also had the highest test parameters on external validation (AUROC range of 0.85-0.87, and AUPRC range of 0.27-0.54) across all time windows. XGBoost without imputation outperformed all models with higher precision and recall (mean difference in AUROC of 0.04; mean difference in AUPRC of 0.15). Features of creatinine, BUN, and red cell distribution width were major drivers of the model's prediction. Conclusions An XGBoost model without imputation for prediction of a composite outcome of either death or dialysis in patients positive for COVID-19 had the best performance, as compared with standard and other machine learning models. KW - COVID-19 KW - dialysis KW - machine learning KW - prediction KW - AKI Y1 - 2021 U6 - https://doi.org/10.2215/CJN.17311120 SN - 1555-9041 SN - 1555-905X VL - 16 IS - 8 SP - 1158 EP - 1168 PB - American Society of Nephrology CY - Washington ER - TY - JOUR A1 - Ebers, Martin A1 - Hoch, Veronica R. S. A1 - Rosenkranz, Frank A1 - Ruschemeier, Hannah A1 - Steinrötter, Björn T1 - The European Commission’s proposal for an Artificial Intelligence Act BT - a critical assessment by members of the Robotics and AI Law Society (RAILS) JF - J : multidisciplinary scientific journal N2 - On 21 April 2021, the European Commission presented its long-awaited proposal for a Regulation “laying down harmonized rules on Artificial Intelligence”, the so-called “Artificial Intelligence Act” (AIA). This article takes a critical look at the proposed regulation. After an introduction (1), the paper analyzes the unclear preemptive effect of the AIA and EU competences (2), the scope of application (3), the prohibited uses of Artificial Intelligence (AI) (4), the provisions on high-risk AI systems (5), the obligations of providers and users (6), the requirements for AI systems with limited risks (7), the enforcement system (8), the relationship of the AIA with the existing legal framework (9), and the regulatory gaps (10). The last section draws some final conclusions (11). KW - artificial intelligence KW - machine learning KW - European Union KW - regulation KW - harmonization KW - Artificial Intelligence Act Y1 - 2021 U6 - https://doi.org/10.3390/j4040043 SN - 2571-8800 VL - 4 IS - 4 SP - 589 EP - 603 PB - MDPI CY - Basel ER - TY - GEN A1 - Panzer, Marcel A1 - Bender, Benedict A1 - Gronau, Norbert T1 - Deep reinforcement learning in production planning and control BT - A systematic literature review T2 - Zweitveröffentlichungen der Universität Potsdam : Wirtschafts- und Sozialwissenschaftliche Reihe N2 - Increasingly fast development cycles and individualized products pose major challenges for today's smart production systems in times of industry 4.0. The systems must be flexible and continuously adapt to changing conditions while still guaranteeing high throughputs and robustness against external disruptions. Deep reinforcement learning (RL) algorithms, which already reached impressive success with Google DeepMind's AlphaGo, are increasingly transferred to production systems to meet related requirements. Unlike supervised and unsupervised machine learning techniques, deep RL algorithms learn based on recently collected sensorand process-data in direct interaction with the environment and are able to perform decisions in real-time. As such, deep RL algorithms seem promising given their potential to provide decision support in complex environments, as production systems, and simultaneously adapt to changing circumstances. While different use-cases for deep RL emerged, a structured overview and integration of findings on their application are missing. To address this gap, this contribution provides a systematic literature review of existing deep RL applications in the field of production planning and control as well as production logistics. From a performance perspective, it became evident that deep RL can beat heuristics significantly in their overall performance and provides superior solutions to various industrial use-cases. Nevertheless, safety and reliability concerns must be overcome before the widespread use of deep RL is possible which presumes more intensive testing of deep RL in real world applications besides the already ongoing intensive simulations. T3 - Zweitveröffentlichungen der Universität Potsdam : Wirtschafts- und Sozialwissenschaftliche Reihe - 198 KW - deep reinforcement learning KW - machine learning KW - production planning KW - production control KW - systematic literature review Y1 - 2021 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-605722 SN - 2701-6277 SN - 1867-5808 ER - TY - CHAP A1 - Panzer, Marcel A1 - Bender, Benedict A1 - Gronau, Norbert T1 - Deep reinforcement learning in production planning and control BT - A systematic literature review T2 - Proceedings of the Conference on Production Systems and Logistics N2 - Increasingly fast development cycles and individualized products pose major challenges for today's smart production systems in times of industry 4.0. The systems must be flexible and continuously adapt to changing conditions while still guaranteeing high throughputs and robustness against external disruptions. Deep rein- forcement learning (RL) algorithms, which already reached impressive success with Google DeepMind's AlphaGo, are increasingly transferred to production systems to meet related requirements. Unlike supervised and unsupervised machine learning techniques, deep RL algorithms learn based on recently collected sensor- and process-data in direct interaction with the environment and are able to perform decisions in real-time. As such, deep RL algorithms seem promising given their potential to provide decision support in complex environments, as production systems, and simultaneously adapt to changing circumstances. While different use-cases for deep RL emerged, a structured overview and integration of findings on their application are missing. To address this gap, this contribution provides a systematic literature review of existing deep RL applications in the field of production planning and control as well as production logistics. From a performance perspective, it became evident that deep RL can beat heuristics significantly in their overall performance and provides superior solutions to various industrial use-cases. Nevertheless, safety and reliability concerns must be overcome before the widespread use of deep RL is possible which presumes more intensive testing of deep RL in real world applications besides the already ongoing intensive simulations. KW - deep reinforcement learning KW - machine learning KW - production planning KW - production control KW - systematic literature review Y1 - 2021 U6 - https://doi.org/10.15488/11238 SN - 2701-6277 SP - 535 EP - 545 PB - publish-Ing. CY - Hannover ER -