05 Fakultät Informatik, Elektrotechnik und Informationstechnik

Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6

Browse

Filters

20227

Settings

Institute: search.filters.UBSInstitut.Institut für Parallele und Verteilte SystemePublication Type: search.filters.UBSTyp.ZeitschriftenartikelStart date: 2022End date: 2022

Search Results

Now showing 1 - 7 of 7
  • Thumbnail Image
    ItemOpen Access
    Enhancing quasi-newton acceleration for fluid-structure interaction
    (2022) Davis, Kyle; Schulte, Miriam; Uekermann, Benjamin
    We propose two enhancements of quasi-Newton methods used to accelerate coupling iterations for partitioned fluid-structure interaction. Quasi-Newton methods have been established as flexible, yet robust, efficient and accurate coupling methods of multi-physics simulations in general. The coupling library preCICE provides several variants, the so-called IQN-ILS method being the most commonly used. It uses input and output differences of the coupled solvers collected in previous iterations and time steps to approximate Newton iterations. To make quasi-Newton methods both applicable for parallel coupling (where these differences contain data from different physical fields) and to provide a robust approach for re-using information, a combination of information filtering and scaling for the different physical fields is typically required. This leads to good convergence, but increases the cost per iteration. We propose two new approaches - pre-scaling weight monitoring and a new, so-called QR3 filter, to substantially improve runtime while not affecting convergence quality. We evaluate these for a variety of fluid-structure interaction examples. Results show that we achieve drastic speedups for the pure quasi-Newton update steps. In the future, we intend to apply the methods also to volume-coupled scenarios, where these gains can be decisive for the feasibility of the coupling approach.
  • Thumbnail Image
    ItemOpen Access
    SMARTEN : a sample-based approach towards privacy-friendly data refinement
    (2022) Stach, Christoph; Behringer, Michael; Bräcker, Julia; Gritti, Clémentine; Mitschang, Bernhard
    Two factors are crucial for the effective operation of modern-day smart services: Initially, IoT-enabled technologies have to capture and combine huge amounts of data on data subjects. Then, all these data have to be processed exhaustively by means of techniques from the area of big data analytics. With regard to the latter, thorough data refinement in terms of data cleansing and data transformation is the decisive cornerstone. Studies show that data refinement reaches its full potential only by involving domain experts in the process. However, this means that these experts need full insight into the data in order to be able to identify and resolve any issues therein, e.g., by correcting or removing inaccurate, incorrect, or irrelevant data records. In particular for sensitive data (e.g., private data or confidential data), this poses a problem, since these data are thereby disclosed to third parties such as domain experts. To this end, we introduce SMARTEN, a sample-based approach towards privacy-friendly data refinement to smarten up big data analytics and smart services. SMARTEN applies a revised data refinement process that fully involves domain experts in data pre-processing but does not expose any sensitive data to them or any other third-party. To achieve this, domain experts obtain a representative sample of the entire data set that meets all privacy policies and confidentiality guidelines. Based on this sample, domain experts define data cleaning and transformation steps. Subsequently, these steps are converted into executable data refinement rules and applied to the entire data set. Domain experts can request further samples and define further rules until the data quality required for the intended use case is reached. Evaluation results confirm that our approach is effective in terms of both data quality and data privacy.
  • Thumbnail Image
    ItemOpen Access
    Effective or predatory funding? : evaluating the hidden costs of grant applications
    (2022) Dresler, Martin; Buddeberg, Eva; Endesfelder, Ulrike; Haaker, Jan; Hof, Christian; Kretschmer, Robert; Pflüger, Dirk; Schmidt, Fabian
    Researchers are spending an increasing fraction of their time on applying for funding; however, the current funding system has considerable deficiencies in reliably evaluating the merit of research proposals, despite extensive efforts on the sides of applicants, grant reviewers and decision committees. For some funding schemes, the systemic costs of the application process as a whole can even outweigh the granted resources - a phenomenon that could be considered as predatory funding. We present five recommendations to remedy this unsatisfactory situation.
  • Thumbnail Image
    ItemOpen Access
    Metrics and algorithms for locally fair and accurate classifications using ensembles
    (2022) Lässig, Nico; Oppold, Sarah; Herschel, Melanie
    To obtain accurate predictions of classifiers, model ensembles comprising multiple trained machine learning models are nowadays used. In particular, dynamic model ensembles pick the most accurate model for each query object, by applying the model that performed best on similar data. Dynamic model ensembles may however suffer, similarly to single machine learning models, from bias, which can eventually lead to unfair treatment of certain groups of a general population. To mitigate unfair classification, recent work has thus proposed fair model ensembles , that instead of focusing (solely) on accuracy also optimize global fairness . While such global fairness globally minimizes bias, imbalances may persist in different regions of the data, e.g., caused by some local bias maxima leading to local unfairness . Therefore, we extend our previous work by including a framework that bridges the gap between dynamic model ensembles and fair model ensembles. More precisely, we investigate the problem of devising locally fair and accurate dynamic model ensembles, which ultimately optimize for equal opportunity of similar subjects. We propose a general framework to perform this task and present several algorithms implementing the framework components. In this paper we also present a runtime-efficient framework adaptation that keeps the quality of the results on a similar level. Furthermore, new fairness metrics are presented as well as detailed informations about necessary data preparations. Our evaluation of the framework implementations and metrics shows that our approach outperforms the state-of-the art for different types and degrees of bias present in training data in terms of both local and global fairness, while reaching comparable accuracy.
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    Protecting sensitive data in the information age : state of the art and future prospects
    (2022) Stach, Christoph; Gritti, Clémentine; Bräcker, Julia; Behringer, Michael; Mitschang, Bernhard
    The present information age is characterized by an ever-increasing digitalization. Smart devices quantify our entire lives. These collected data provide the foundation for data-driven services called smart services. They are able to adapt to a given context and thus tailor their functionalities to the user’s needs. It is therefore not surprising that their main resource, namely data, is nowadays a valuable commodity that can also be traded. However, this trend does not only have positive sides, as the gathered data reveal a lot of information about various data subjects. To prevent uncontrolled insights into private or confidential matters, data protection laws restrict the processing of sensitive data. One key factor in this regard is user-friendly privacy mechanisms. In this paper, we therefore assess current state-of-the-art privacy mechanisms. To this end, we initially identify forms of data processing applied by smart services. We then discuss privacy mechanisms suited for these use cases. Our findings reveal that current state-of-the-art privacy mechanisms provide good protection in principle, but there is no compelling one-size-fits-all privacy approach. This leads to further questions regarding the practicality of these mechanisms, which we present in the form of seven thought-provoking propositions.
  • Thumbnail Image
    ItemOpen Access
    CLAIRE : parallelized diffeomorphic image registration for large-scale biomedical imaging applications
    (2022) Himthani, Naveen; Brunn, Malte; Kim, Jae-Youn; Schulte, Miriam; Mang, Andreas; Biros, George
    We study the performance of CLAIRE - a diffeomorphic multi-node, multi-GPU image-registration algorithm and software-in large-scale biomedical imaging applications with billions of voxels. At such resolutions, most existing software packages for diffeomorphic image registration are prohibitively expensive. As a result, practitioners first significantly downsample the original images and then register them using existing tools. Our main contribution is an extensive analysis of the impact of downsampling on registration performance. We study this impact by comparing full-resolution registrations obtained with CLAIRE to lower resolution registrations for synthetic and real-world imaging datasets. Our results suggest that registration at full resolution can yield a superior registration quality-but not always. For example, downsampling a synthetic image from 10243 to 2563 decreases the Dice coefficient from 92% to 79%. However, the differences are less pronounced for noisy or low contrast high resolution images. CLAIRE allows us not only to register images of clinically relevant size in a few seconds but also to register images at unprecedented resolution in reasonable time. The highest resolution considered are CLARITY images of size 2816×3016×1162. To the best of our knowledge, this is the first study on image registration quality at such resolutions.