Logo des Repositoriums

Auflistung nach Autor:in "Eleks, Marian"

1 - 3 von 3
  • Konferenzbeitrag
    Privacy Aware Processing
    (INFORMATIK 2023 - Designing Futures: Zukünfte gestalten, 2023) Eleks, Marian; Rebstadt, Jonas; Kortum, Henrik; Thomas, Oliver
    In many machine learning (ML) applications, the provision of data and the training as well as the analysis of machine learning systems are performed by distinct actors, a data owner and a data consumer. To protect sensitive information in these ML-scenarios, privacy aware machine learning (PAML) methods are often applied to the data before sharing. Based on the type of PAML methods used, data understanding and preparation as defined in the CRISP-DM model become more difficult if not impossible. To enable these steps, we propose a method to share a variety of uncritical information with the data consumer who is then able to define the necessary processing steps on a meta-level. These are then applied to the data in the data owners local trusted environment before the PAML-methods whereupon the prepared and protected data is shared.
  • Konferenzbeitrag
    Privacy, Utility, Effort, Transparency and Fairness: Identifying and Swaying Trade-offs in Privacy Preserving Machine Learning through Hybrid Methods
    (INFORMATIK 2024, 2024) Eleks, Marian; Ihler Jakob; Rebstadt, Jonas; Kortum-Landwehr, Henrik; Thomas, Oliver
    As Artificial Intelligence (AI) permeates most economic sectors, the discipline Privacy Preserving Machine Learning (PPML) gains increasing importance as a way to ensure appropriate handling of sensitive data in the machine learning process. Although PPML-methods stand to provide privacy protection in AI use cases, each one comes with a trade-off. Practitioners applying PPML-methods increasingly request an overview of the types and impacts of these trade-offs. To aid this gap in knowledge, this article applies design science research to collect trade-off dimensions and method impacts in an extensive literature review. It then evaluates the specific trade-offs with a focus group of experts and finally constructs an overview over PPML-methods and method combinations’ impact. The final trade-off dimensions are privacy, utility, effort, transparency, and fairness. Seven PPML-methods and their combinations are evaluated according to their impact in these dimensions, resulting in a vast collection of design knowledge and identified research gaps.
  • Zeitschriftenartikel
    SECAI – Sustainable Heating through Edge-Cloud-based AI Systems
    (HMD Praxis der Wirtschaftsinformatik: Vol. 60, No. 4, 2023) Kortum, Henrik; Hagen, Simon; Eleks, Marian; Rebstadt, Jonas; Remark, Florian; Lowin, Maximilian; Wilson, Cristina Mihale; Eberhardt, Birgid; Roß, Andree; Maihöfner, Dominik; Hinz, Oliver; Thomas, Oliver
    Etwa 18 % der CO 2 -Emissionen in Deutschland entstehen durch die Beheizung, Kühlung und Warmwasserbereitstellung von Gebäuden, wobei mehr als 75 % der deutschen Haushalte fossile Brennstoffe wie Erdgas und Erdöl nutzen. Der in dieser Arbeit vorgestellte SECAI ( S ustainable heating through E dge- C loud-based A rtificial I ntelligence Systems)-Ansatz verfolgt das Ziel, die Heizungssteuerung in Mehrfamilienhäusern und damit den CO 2 -Verbrauch durch den Einsatz von Informationstechnologien zu reduzieren. Der SECAI-Ansatz betrachtet dabei das gesamte Ökosystem bestehend aus Sensoren, Einzelraumregelungen, Zentralheizung, Mietenden und Vermietenden. Dabei wird der Heizbedarf von Privatwohnungen KI-basiert analysiert, um darauf aufbauend optimierte und abgestimmte Heizpläne für Gebäudekomplexe und Wohnungen zu erstellen, die in der Lage sind, durch Edge-Cloud-Technologien, Sensorik und Federated Learning ad hoc und datenschutzkonform auf Änderungen im Nutzungsverhalten zu reagieren. Diese Informationen werden zudem für die KI-basierte Steuerung der zentralen Heizanlagen im Gebäude verwendet, in denen Wärme und Warmwasser für alle Wohnungen erzeugt wird. Hierfür betrachtet SECAI vier Ebenen. Diese reichen von Sensoren und Aktoren (Nano), über die Wohnung (Mikro) und das Gebäude (Meso) bis zu Gebäudekomplexen und gleicharten Gebäuden (Makro) und stehen bei der Beheizung in starker Abhängigkeit zueinander. Rund um die SECAI-Lösung entsteht dabei ein komplexes Ökosystem in dem Mietende, die Wohnungswirtschaft, Heizungshersteller und Anbieter von IoT-Lösungen mit Produkten und Diensten in Interaktion treten. Approximately 18% of CO2 emissions in Germany are caused by the heating, cooling and hot water supply of buildings, with more than 75% of households using fossil fuels such as natural gas and oil. The SECAI (Sustainable heating through Edge-Cloud-based Artificial Intelligence Systems) approach presented in this paper aims to reduce heating control in multi-residential buildings, and thus CO2 consumption, through the use of information technology. The SECAI approach considers the entire ecosystem consisting of sensors, individual room controls, central heating, tenants and landlords. This involves an AI-based analysis of the heating requirements of private apartments, based on which optimized and coordinated heating plans can be created for building complexes. Edge cloud technologies, sensor technology and federated learning enable these plans to react ad hoc and in compliance with data protection regulations to changes in usage behavior. The information is also used for AI-based control of the central heating systems within the building, where heating and hot water are generated for all apartments. For this purpose, SECAI considers four layers. These range from sensors and actuators (nano), to the apartment (micro), to the building (meso), to building complexes and same-type buildings (macro), and are highly interdependent. A complex ecosystem is being created around the SECAI solution in which tenants, the housing industry, heating manufacturers and providers of IoT solutions interact with products and services.