05 Fakultät Informatik, Elektrotechnik und Informationstechnik
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6
Browse
17 results
Search Results
Item Open Access Exploiting domain knowledge to address class imbalance and a heterogeneous feature space in multi-class classification(2023) Hirsch, Vitali; Reimann, Peter; Treder-Tschechlov, Dennis; Schwarz, Holger; Mitschang, BernhardReal-world data of multi-class classification tasks often show complex data characteristics that lead to a reduced classification performance. Major analytical challenges are a high degree of multi-class imbalance within data and a heterogeneous feature space, which increases the number and complexity of class patterns. Existing solutions to classification or data pre-processing only address one of these two challenges in isolation. We propose a novel classification approach that explicitly addresses both challenges of multi-class imbalance and heterogeneous feature space together. As main contribution, this approach exploits domain knowledge in terms of a taxonomy to systematically prepare the training data. Based on an experimental evaluation on both real-world data and several synthetically generated data sets, we show that our approach outperforms any other classification technique in terms of accuracy. Furthermore, it entails considerable practical benefits in real-world use cases, e.g., it reduces rework required in the area of product quality control.Item Open Access Metrics and algorithms for locally fair and accurate classifications using ensembles(2022) Lässig, Nico; Oppold, Sarah; Herschel, MelanieTo 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.Item Open Access Enhancing quasi-newton acceleration for fluid-structure interaction(2022) Davis, Kyle; Schulte, Miriam; Uekermann, BenjaminWe 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.Item Open Access Research data management in simulation science : infrastructure, tools, and applications(2024) Flemisch, Bernd; Hermann, Sibylle; Herschel, Melanie; Pflüger, Dirk; Pleiss, Jürgen; Range, Jan; Roy, Sarbani; Takamoto, Makoto; Uekermann, BenjaminResearch Data Management (RDM) has gained significant traction in recent years, being essential to allowing research data to be, e.g., findable, accessible, interoperable, and reproducible (FAIR), thereby fostering collaboration or accelerating scientific findings. We present solutions for RDM developed within the DFG-Funded Cluster of Excellence EXC2075 Data-Integrated Simulation Science (SimTech). After an introduction to the scientific context and challenges faced by simulation scientists, we outline the general data management infrastructure and present tools that address these challenges. Exemplary domain applications demonstrate the use and benefits of the proposed data management software solutions. These are complemented by additional measures for enablement and dissemination to foster the adoption of these techniques.Item Open Access Editorial - special issue on security and privacy in blockchains and the IoT volume II(2023) Stach, Christoph; Gritti, ClémentineItem Open Access SMARTEN : a sample-based approach towards privacy-friendly data refinement(2022) Stach, Christoph; Behringer, Michael; Bräcker, Julia; Gritti, Clémentine; Mitschang, BernhardTwo 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.Item Open Access Enhancing data trustworthiness in explorative analysis : an interactive approach for data quality monitoring(2024) Behringer, Michael; Hirmer, Pascal; Villanueva, Alejandro; Rapp, Jannis; Mitschang, BernhardThe volume of data to be analyzed has increased tremendously in recent years. In order to extract knowledge from this data, domain experts gain new insights with the help of graphical analysis tools for explorative analyses. Here, the reliability and trustworthiness of an exploratory analysis is determined by the quality of the underlying data. Existing approaches require manual testing to ensure data quality which is often neglected. This research aims to introduce a novel interactive approach for seamlessly integrating data quality considerations into the process of explorative data analysis conducted by domain experts. We derive requirements, conduct an extensive literature review, and develop an approach that efficiently combines stakeholders’ strengths, allowing unobtrusive data quality integration in interactive analysis. Our approach enhances trustworthiness due to unobtrusive monitoring of data quality within the context of explorative data analysis. Domain experts gain insights more reliably, bridging the gap between technical requirements and domain expertise. In conclusion, our research presents a promising solution for improving the reliability and trustworthiness of explorative data analysis, especially for domain experts who may lack technical knowledge. By seamlessly integrating data quality into the analytical process, we empower domain experts to extract valuable insights from the ever-increasing volume of data, thereby advancing the field of data-driven decision-making.Item Open Access The lakehouse : state of the art on concepts and technologies(2024) Schneider, Jan; Gröger, Christoph; Lutsch, Arnold; Schwarz, Holger; Mitschang, BernhardIn the context of data analytics, so-called lakehouses refer to novel variants of data platforms that attempt to combine characteristics of data warehouses and data lakes. In this way, lakehouses promise to simplify enterprise analytics architectures, which often suffer from high operational costs, slow analytical processes and further shortcomings resulting from data replication. However, different views and notions on the lakehouse paradigm exist, which are commonly driven by individual technologies and varying analytical use cases. Therefore, it remains unclear what challenges lakehouses address, how they can be characterized and which technologies can be leveraged to implement them. This paper addresses these issues by providing an extensive overview of concepts and technologies that are related to the lakehouse paradigm and by outlining lakehouses as a distinct architectural approach for data platforms. Concepts and technologies from literature with regard to lakehouses are discussed, based on which a conceptual foundation for lakehouses is established. In addition, several popular technologies are evaluated regarding their suitability for the building of lakehouses. All findings are supported and demonstrated with the help of a representative analytics scenario. Typical challenges of conventional data platforms are identified, a new, sharper definition for lakehouses is proposed and technical requirements for lakehouses are derived. As part of an evaluation, these requirements are applied to several popular technologies, of which frameworks for data lakes turn out to be particularly helpful for the construction of lakehouses. Our work provides an overview of the state of the art and a conceptual foundation for the lakehouse paradigm, which can support future research.Item Open Access Data is the new oil - sort of : a view on why this comparison is misleading and its implications for modern data administration(2023) Stach, ChristophCurrently, data are often referred to as the oil of the 21st century. This comparison is not only used to express that the resource data are just as important for the fourth industrial revolution as oil was for the technological revolution in the late 19th century. There are also further similarities between these two valuable resources in terms of their handling. Both must first be discovered and extracted from their sources. Then, the raw materials must be cleaned, preprocessed, and stored before they can finally be delivered to consumers. Despite these undeniable similarities, however, there are significant differences between oil and data in all of these processing steps, making data a resource that is considerably more challenging to handle. For instance, data sources, as well as the data themselves, are heterogeneous, which means there is no one-size-fits-all data acquisition solution. Furthermore, data can be distorted by the source or by third parties without being noticed, which affects both quality and usability. Unlike oil, there is also no uniform refinement process for data, as data preparation should be tailored to the subsequent consumers and their intended use cases. With regard to storage, it has to be taken into account that data are not consumed when they are processed or delivered to consumers, which means that the data volume that has to be managed is constantly growing. Finally, data may be subject to special constraints in terms of distribution, which may entail individual delivery plans depending on the customer and their intended purposes. Overall, it can be concluded that innovative approaches are needed for handling the resource data that address these inherent challenges. In this paper, we therefore study and discuss the relevant characteristics of data making them such a challenging resource to handle. In order to enable appropriate data provisioning, we introduce a holistic research concept from data source to data sink that respects the processing requirements of data producers as well as the quality requirements of data consumers and, moreover, ensures a trustworthy data administration.Item Open Access Proximity-based service discovery for distributed digital twin systems(2025) Rothermel, Kurt; Herzog, Otthein; Wu, Zhiqiang SiegfriedOver the past decade, there has been a significant increase in interest in digital twin (DT) technology in a variety of domains. While research on DTs of single assets was initially prevalent, there has been a notable shift towards distributed systems of DTs, which connect to each other to collaborate. Typically, collaboration is enabled by DTs providing services that can be consumed by other DTs. In service-oriented systems, a service is typically identified by type information. However, this is not sufficient in distributed DT systems, where DTs associated with different physical entities may provide the same type of service. Consequently, selecting the appropriate service depends not only on the service type, but also on the associated physical entity. However, requiring DTs to know the mapping of services to their physical environment is not feasible for large dynamic systems. This paper presents a novel proximity-based service discovery method that allows DTs to select services based on service type and their proximity to other objects. That is, service specifications are fully abstracted from the mapping of services to physical objects, relieving DTs from maintaining information about this mapping. Furthermore, service discovery is robust to changes in the physical environment and service population. The proposed service discovery method has been implemented on top of a spatial DBMS. We argue that this implementation is optimal in terms of network utilization and latency, and perform comprehensive evaluations to show the performance of discovery queries as a function of their complexity.