DigitalAngewandte InformatikBeitrag (Sammelband oder Tagungsband)
Patrick Glauner
Unlocking the Power of Artificial Intelligence for Your Business
Innovative Technologies for Market Leadership: Investing in the Future
2020
ISBN: 978-3-030-41308-8
DigitalAngewandte InformatikTC FreyungBeitrag (Sammelband oder Tagungsband)
C. Hoermann, Raphaela Pagany, K. Kirchner, Wolfgang Dorner, M. Heurich, I. Storch
Predicting the risk of deer-vehicle collisions by inferring rules learnt from deer experience and movement patterns in the vicinity of roads
Proceedings of the 2020 10th International Conference on Advanced Computer Information Technologies (ACIT) [September 6-8, 2020; Deggendorf]
2020
DOI: 10.1109/ACIT49673.2020.9208843
Abstract anzeigen
Estimates of annual deer-vehicle collisions exceed one million incidences in Europe. Consequently, we were analyzing whether an animal’s experience and movement pattern close to roads can provide crucial information for accident prevention and mitigation measures. We applied an innovative approach using machine learning and step selection analyses to find rules and patterns in deer movement data for a better understanding of the spatio-temporal dynamics in wildlife-vehicle collisions. The rule tree indicated highest collision probabilities when the mean distance to a road of a roe deer tracking path was shorter than 192 meters and the roe deer crossed in more unfamiliar areas of its home range. The step selection function analysis revealed no obvious road avoidance and more road crossings in areas with less understory vegetation. Our results demonstrate the power of learned threshold values and step selection functions modelling results for a better understanding of the factors driving deer behavior in the vicinity of roads.
DigitalAngewandte InformatikTC FreyungBeitrag (Sammelband oder Tagungsband)
Raphaela Pagany, Javier Valdés, Wolfgang Dorner
Risk prediction of wildlife-vehicle collisions comparing machine learning methods and data use
Proceedings of the 2020 10th International Conference on Advanced Computer Information Technologies (ACIT) [September 6-8, 2020; Deggendorf]
2020
DOI: 10.1109/ACIT49673.2020.9208946
Abstract anzeigen
Risk prediction of wildlife-vehicle collisions is crucial for reducing the abundant number of road accidents with wildlife worldwide. In this study, three different machine learning approaches - Gaussian Naive Bayes, Stochastic Gradient Descent, Random Forest - relying on recorded accident data, environmental, temporal and infrastructural parameters, are used to test the impurity of these parameters, and to predict the risk of wildlife-vehicle collisions. We use a dataset of one million police recorded accidents for southeastern Bavaria. Based on different approaches for feature engineering, we show the importance of data pre-processing and cleaning. With Random Forest, we receive a prediction accuracy of accidents of 86.7%. Additionally, the transferability of the three approaches is discussed, and possible ways of dynamic warnings via smartphone are presented.
DigitalAngewandte InformatikTC FreyungZeitschriftenartikel
Raphaela Pagany
Wildlife-vehicle collisions—Influencing factors, data collection and research methods
Biological Conservation, vol. 251, no. November
2020
DOI: 10.1016/j.biocon.2020.108758
Abstract anzeigen
Wildlife-vehicle collisions (WVCs) are caused by the close interaction of human and wildlife habitats worldwide. The large number of globally distributed accidents and the variety of environmental impacts characterize WVCs as intricate and challenging to predict. However, numerous research studies have been conducted to understand the causal relationships between drivers, animals, and the environment. In this paper, 645 publications are reviewed to provide an overview and a wide-ranging knowledge about WVC research. The study gathers the influencing factors on WVCs, systematizes the approaches for data collection, and identifies the main developments in analysis and predicting methods for WVCs. Factors such as the proximity to forest, a gentle topography with sparsely curves, street width, and seasonal differences are common denominators for WVCs - independent of the species -, while traffic volume, the distance to urban areas, or road accompanying infrastructure are not clearly assignable influencing or non-influencing factors. Different ways of data collection are observed, which range from carcass surveys by ecologists or crowdsourcing for species conservation to nearly real-time official reporting by involved parties as a basis for driving safety. Data collection and quality are discussed for their applicability, in particular, regarding the currently used analysing approaches for WVCs. Additionally, the advantages of the rarely employed machine learning approaches are discussed in terms of dynamic WVC risk prediction - including large-scale and temporally unrestricted transferability. These approaches may be helpful for prospective warning and road safety management on a global scale.
DigitalAngewandte InformatikTC GrafenauZeitschriftenartikel
Ali Fallah Tehrani, M. Strickert, Diane Ahrens
Class of Monotone Kernelized Classifiers on the basis of the Choquet Integral
Expert Systems, vol. 37, no. First published: 21 January 2020, pp. 1-15
2020
DOI: 10.1111/exsy.12506
Abstract anzeigen
The key property of monotone classifiers is that increasing (decreasing) input values lead to increasing (decreasing) the output value. Preserving monotonicity for a classifier typically requires many constraints to be respected by modeling approaches such as artificial intelligence techniques. The type of constraints strongly depends on the modeling assumptions. Of course, for sophisticated models such conditions might be very complex. In this study we present a new family of kernels that we call it Choquet kernels. Henceforth it allows for employing popular kernel‐based methods such as support vector machines. Instead of a naïve approach with exponential computational complexity we propose an equivalent formulation with quadratic time in the number of attributes. Furthermore, since coefficients derived from kernel solutions are not necessarily monotone in the dual form, different approaches are proposed to monotonize coefficients. Finally experiments illustrate beneficial properties of the Choquet kernels.
DigitalAngewandte InformatikAngewandte WirtschaftswissenschaftenZeitschriftenartikel
Marco Kretschmann, Andreas Fischer, Benedikt Elser
Extracting Keywords from Publication Abstracts for an Automated Researcher Recommendation System
Digitale Welt (Proceedings of the First International Symposium on Applied Artificial Intelligence in Conjunction with DIGICON), vol. 4, no. 1, pp. 20-25
2020
DOI: 10.1007/s42354-019-0227-2
Abstract anzeigen
This paper presents an automated keyword assignment system for scientific abstracts. That system is applied to paper abstracts collected in a local publication database and used to drive a researcher recommendation system. Problems like low data volume and missing keywords are discussed. For remediation, training is performed on an extended data set based on large online publication databases. Additionally a closer look at label imbalance in the dataset is taken. Ten multi-label classification algorithms for assigning keywords from a given catalogue to a scientific abstract are compared. The usage of binary relevance as transformation method with LightGBM as classifier yields the best results. Random oversampling before the training phase additionally increases the F1-Score by around 5-6%.
DigitalAngewandte InformatikBeitrag (Sammelband oder Tagungsband)
Monica I. Ciolacu, Leon Binder, P. M. Svasta, D. Stoichescu, I. Tache
Education 4.0 - Jump to Innovation IoT in Higher Education
Proceedings of the 2019 IEEE 25th International Symposium for Design and Technology in Electronic Packaging (SIITME) [Oct 23-26, 2019; Cluj-Napoca, Romania], New York, NY, USA
2020
DOI: 10.1109/SIITME47687.2019.8990825
Abstract anzeigen
Artificial Intelligence (AI) will play a key role in Higher Education. Our contribution leads the road to innovation within IoT for Education 4.0, using a system comprising smartwatch data, health data, learning analytics and Artificial Intelligence. By using embedded sensors from wearable devices, we can add value to Education 4.0. Smartwatches are IoT devices, equipped with a multitude of embedded sensors collecting distraction-free huge amounts of real-time data during students’ learning. In our paper we highlight advantages of wrist-based wearables like smartwatches for Education 4.0. We develop a User Experience Questionnaire for the measurement of acceptance of advanced electronic technology in Higher Education. We identify as experiment’s results the most important sensors and protocols for Education 4.0.
DigitalAngewandte InformatikTC GrafenauBeitrag (Sammelband oder Tagungsband)
Monica I. Ciolacu, Leon Binder, Heribert Popp
Enabling IoT in Education 4.0 with Biosensors from Wearables and Artificial Intelligence
Proceedings of the 2019 IEEE 25th International Symposium for Design and Technology in Electronic Packaging (SIITME) [Oct 23-26, 2019; Cluj-Napoca, Romania], New York, NY, USA
2020
DOI: 10.1109/SIITME47687.2019.8990763
Abstract anzeigen
A major challenge for Education 4.0 is to make use of wearable devices for helping students in monitoring their learning behavior and their activities (steps, heart rate variability, and heart rate) in real-time. The first aim of this paper is to present our implementation of adaptivity and Artificial Intelligence (AI) methods within the Education 4.0 process. In this work, we investigate embedded biosensors (noninvasive, low-cost, and distraction-free) used in smartphones and smartwatches. The next objective is to enable IoT for Higher Education, i.e. a novel system assisted by AI that takes embedded biosensor data and environmental data into account in order to estimate students’ wellbeing and health. In this regard, we propose a framework that uses wearable devices to collect data with biofeedback methods to support students’ academic success.
DigitalAngewandte InformatikVortrag
Monica I. Ciolacu
Education 4.0: Learning Analytics in der Lehre mit Künstliche Intelligenz und Sensoren
60-minütiger Workshop (Focus Group I: KI in der Bildung - Learning Analytics)
Leadership & Innovation Talk 2019: Do you trust this Robot? - Künstliche Intelligenz fordert uns heraus, München
2019
NachhaltigAngewandte InformatikTC FreyungZeitschriftenartikel
D. Kammerl, Roland Zink
Nachhaltigkeitsdiamant – Bewertungs- und Implementierungsmethode für eine nachhaltigkeitsorientierte Produktentwicklung
Bavarian Journal of Applied Sciences, vol. 5, no. 1, pp. 423-434
2019
DOI: 10.25929/bjas.v5i1.71
Abstract anzeigen
Im vorliegenden Beitrag wird eine Methode beschrieben, um Nachhaltigkeit in der Produkt- und Produkt-Service-System-Entwicklung zu bewerten. Der Fokus liegt dabei insbesondere auf der frühen Phase der Entwicklung, um im Gegensatz zu einer Vielzahl der bestehenden Bewertungsmethoden proaktiv und nicht reaktiv handeln zu können. Mithilfe der Methode wird der komplexe Sachverhalt der Nachhaltigkeit messbar und graphisch auf einfach verständliche Art und Weise dargestellt.
Version: 2.1.2-qf2