Refine
Year of publication
Document Type
- Conference Proceeding (29)
- Article (reviewed) (19)
- Article (unreviewed) (17)
- Doctoral Thesis (5)
- Report (4)
- Letter to Editor (1)
- Study Thesis (1)
Conference Type
- Konferenzartikel (23)
- Konferenz-Abstract (3)
- Konferenz-Poster (1)
- Sonstiges (1)
Is part of the Bibliography
- yes (76) (remove)
Keywords
- Deep Leaning (3)
- Deep Learning (3)
- Hochschuldidaktik (3)
- machine learning (3)
- Advanced Footwear Technology (2)
- Alexander von Humboldt (2)
- Artificial Intelligence (2)
- Biomechanics (2)
- Digitaler Zwilling (2)
- Exercise Science (2)
Institute
- Fakultät Maschinenbau und Verfahrenstechnik (M+V) (25)
- Fakultät Elektrotechnik, Medizintechnik und Informatik (EMI) (ab 04/2019) (24)
- INES - Institut für nachhaltige Energiesysteme (13)
- Fakultät Medien (M) (ab 22.04.2021) (12)
- Fakultät Wirtschaft (W) (11)
- IMLA - Institute for Machine Learning and Analytics (9)
- ivESK - Institut für verlässliche Embedded Systems und Kommunikationselektronik (4)
- IBMS - Institute for Advanced Biomechanics and Motion Studies (ab 16.11.2022) (3)
- ACI - Affective and Cognitive Institute (2)
- IfTI - Institute for Trade and Innovation (2)
Open Access
- Diamond (76) (remove)
In recent years, the topic of embedded machine learning has become very popular in AI research. With the help of various compression techniques such as pruning, quantization and others compression techniques, it became possible to run neural networks on embedded devices. These techniques have opened up a whole new application area for machine learning. They range from smart products such as voice assistants to smart sensors that are needed in robotics. Despite the achievements in embedded machine learning, efficient algorithms for training neural networks in constrained domains are still lacking. Training on embedded devices will open up further fields of applications. Efficient training algorithms would enable federated learning on embedded devices, in which the data remains where it was collected, or retraining of neural networks in different domains. In this paper, we summarize techniques that make training on embedded devices possible. We first describe the need and requirements for such algorithms. Then we examine existing techniques that address training in resource-constrained environments as well as techniques that are also suitable for training on embedded devices, such as incremental learning. At the end, we also discuss which problems and open questions still need to be solved in these areas.
Nowadays decarbonisation of the energy system is one of the main concerns for most governments. Renewable energy technologies, such as rooftop photovoltaic systems and home battery storage systems, are changing the energy system to be more decentralised. As a consequence, new ways of energy business models are emerging, e.g., peer-to-peer energy trading. This new concept provides an online marketplace where direct energy exchange can occur between its participants. The purpose of this study is to conduct a content analysis of the existing literature, ongoing research projects, and companies related to peer-to-peer energy trading. From this review, a summary of the most important aspects and journal papers is assessed, discussed, and classified. It was found that the different energy market types were named in various ways and a proposal for standard language for the several peer-to-peer market types and the different actors involved is suggested. Additionally, by grouping the most important attributes from peer-to-peer energy trading projects, an assessment of the entry barrier and scalability potential is performed by using a characterisation matrix.
Featherweight Generic Go (FGG) is a minimal core calculus modeling the essential features of the programming language Go. It includes support for overloaded methods, interface types, structural subtyping and generics. The most straightforward semantic description of the dynamic behavior of FGG programs is to resolve method calls based on runtime type information of the receiver.
This article shows a different approach by defining a type-directed translation from FGG to an untyped lambda-calculus. The translation of an FGG program provides evidence for the availability of methods as additional dictionary parameters, similar to the dictionary-passing approach known from Haskell type classes. Then, method calls can be resolved by a simple lookup of the method definition in the dictionary.
Every program in the image of the translation has the same dynamic semantics as its source FGG program. The proof of this result is based on a syntactic, step-indexed logical relation. The step-index ensures a well-founded definition of the relation in the presence of recursive interface types and recursive methods.
Recent advances in spiked shoe design, characterized by increased longitudinal stiffness, thicker midsole foams, and reconfigured geometry are considered to improve sprint performance. However, so far there is no empirical data on the effects of advanced spikes technology on maximal sprinting speed (MSS) published yet. Consequently, we assessed MSS via ‘flying 30m’ sprints of 44 trained male (PR: 10.32 s - 12.08 s) and female (PR: 11.56 s - 14.18 s) athletes, wearing both traditional and advanced spikes in a randomized, repeated measures design. The results revealed a statistically significant increase in MSS by 1.21% on average when using advanced spikes technology. Notably, 87% of participants showed improved MSS with the use of advanced spikes. A cluster analysis unveiled that athletes with higher MSS may benefit to a greater extent. However, individual responses varied widely, suggesting the influence of multiple factors that need detailed exploration. Therefore, coaches and athletes are advised to interpret the promising performance enhancements cautiously and evaluate the appropriateness of the advanced spike technology for their athletes critically.
Alexander von Humboldt, a German scientist and explorer of the 19th century, viewed the natural world holistically and described the harmony of nature among the diversity of the physical world as a conjoining between all physical disciplines. He noted in his diary: “Everything is interconnectedness.”
The main feature of Humboldt’s pioneering work was later named “Humboldtian science”, meaning the accurate study of interconnected real phenomena in order to find a definite law and a dynamic cause.
Following Humboldt's idea of nature, an Internet edition of his works must preserve the author’s original intention, retain an awareness of all relevant works, and still adhere to the requirements of scholarly edition.
At the present time, however, the highly unconventional form of his publications has undermined the awareness and a comprehensive study of Humboldt’s works.
Digital libraries should supply dynamic links to sources, maps, images, graphs and relevant texts. New forms of interaction and synthesis between humanistic texts and scientific observation need to be created.
Information technology is the only way to do justice to the broad range of visions, descriptions and the idea of nature of Humboldt’s legacy. It finally leads to virtual research environments as an adequate concept to redesign our digital archives, not only for Humboldt’s documents, but for all interconnected data.
Due to its performance, the field of deep learning has gained a lot of attention, with neural networks succeeding in areas like Computer Vision (CV), Neural Language Processing (NLP), and Reinforcement Learning (RL). However, high accuracy comes at a computational cost as larger networks require longer training time and no longer fit onto a single GPU. To reduce training costs, researchers are looking into the dynamics of different optimizers, in order to find ways to make training more efficient. Resource requirements can be limited by reducing model size during training or designing more efficient models that improve accuracy without increasing network size.
This thesis combines eigenvalue computation and high-dimensional loss surface visualization to study different optimizers and deep neural network models. Eigenvectors of different eigenvalues are computed, and the loss landscape and optimizer trajectory are projected onto the plane spanned by those eigenvectors. A new parallelization method for the stochastic Lanczos method is introduced, resulting in faster computation and thus enabling high-resolution videos of the trajectory and secondorder information during neural network training. Additionally, the thesis presents the loss landscape between two minima along with the eigenvalue density spectrum at intermediate points for the first time.
Secondly, this thesis presents a regularization method for Generative Adversarial Networks (GANs) that uses second-order information. The gradient during training is modified by subtracting the eigenvector direction of the biggest eigenvalue, preventing the network from falling into the steepest minima and avoiding mode collapse. The thesis also shows the full eigenvalue density spectra of GANs during training.
Thirdly, this thesis introduces ProxSGD, a proximal algorithm for neural network training that guarantees convergence to a stationary point and unifies multiple popular optimizers. Proximal gradients are used to find a closed-form solution to the problem of training neural networks with smooth and non-smooth regularizations, resulting in better sparsity and more efficient optimization. Experiments show that ProxSGD can find sparser networks while reaching the same accuracy as popular optimizers.
Lastly, this thesis unifies sparsity and neural architecture search (NAS) through the framework of group sparsity. Group sparsity is achieved through ℓ2,1-regularization during training, allowing for filter and operation pruning to reduce model size with minimal sacrifice in accuracy. By grouping multiple operations together, group sparsity can be used for NAS as well. This approach is shown to be more robust while still achieving competitive accuracies compared to state-of-the-art methods
This paper has the objective of creating a framework for a different cultural dimension of corporate entrepreneurship leading to corporate entrepreneurial culture (CEC). The analysis of CEC is based on a review of existing concepts of organisational culture and entrepreneurship. They are combined to create a framework of CEC, including macro- and microlevels and examples of subcultures. Core ideas of the framework are validated by qualitative interviews with ten experts. The identified organisational category of the CEC framework is defined by the levels of micro-cultures or subcultures and includes the upper levels of the hierarchy, including the industry level. Geographic categories such as regional or national culture are also part of the system. The individual category of the CEC framework is characterised by competencies (including aspects such as motivation, creativity, mobilising others, coping with uncertainty, teamwork and social competencies) and entrepreneurial personalities. The results of the interviews show the importance of these individual competencies for a lively CEC. The different levels, such as national and professional cultures, as a dimension of the organisational category of the framework are also confirmed by the interviews. The findings indicate that the individual category of CEC could be used for job satisfaction or engagement and the degree of CEC of an organisation could be defined and developed by the organisational category. The identified framework contributes to an understanding of this complex topic and supports companies in the implementation of entrepreneurial ideas in different organisational contexts.
Auswirkung eines Importstopps russischer Energieträger auf die Klimaschutzziele in Deutschland
(2022)
Ein Importstopp russischer Energieträger nach Deutschland wird derzeit vermehrt diskutiert. Wir wollen die Diskussion unterstützen, indem wir einen Weg zeigen, wie das Elektrizitätssystem in Deutschland kurzfristig mit geringen Energieimporten auskommt und welche Maßnahmen notwendig sind, um die Klimaschutzziele trotzdem einzuhalten. Die Ergebnisse eines solchen Energiewendeszenarios mit reduzierter Importabhängigkeit werden mit dem Energiesystemmodell MyPyPSA-Ger berechnet. Die wichtigsten Erkenntnisse sind, dass ein zügiger Ausbau Erneuerbarer Energien und von Speichertechnologien • die Abhängigkeit des deutschen Elektrizitätssystems von Energieimporten deutlich reduziert. • auch langfristig keine wesentlichen Importe der Energieträger Erdgas, Steinkohle und Mineralöl nach sich zieht. • über die Klimaziele der Bundesregierung hinaus das 1,5-Grad-Ziel im Elektrizitätssystem erreicht wird.
Automatic Identification of Travel Locations in Rare Books - Object Oriented Information Management
(2017)
The digital content of the Internet is growing exponentially and mass digitization of printed media opens access to literature, in particular the genre of travel literature from the 18th and 19th century, which consists of diaries or travel books describing routes, observations or inspirations. The identification of described locations in the digital text is a long-standing challenge which requires information technology to supply dynamic links to sources by new forms of interaction and synthesis between humanistic texts and scientific observations.
Using object oriented information technology, a prototype of a software tool is developed which makes it possible to automatically identify geographic locations and travel routes mentioned in rare books. The information objects contain properties such as names and classification codes for populated places, streams, mountains and regions. Together, with the latitudes and longitudes of every single location, it is possible to geo-reference this information in order that all processed and filtered datasets can be displayed by a map application. This method has already been used in the Humboldt Digital Library to present Alexander von Humboldt’s maps and was tested in a case study to prove the correctness and reliability of the automatic identification of locations based on the work of Alexander von Humboldt and Johann Wolfgang von Goethe.
The results reveal numerous errors due to misspellings, change of location names, equality of terms and location names. But on the other hand it becomes very clear that results of the automatic object detection and recognition can be improved by error-free and comprehensive sources. As a result an increase in quality and usability of the service can be expected, accompanied by more options to detect unknown locations in the descriptions of rare books.