Ongoing collaborations between Paderborn and Bielefeld University:
TRR 318: Constructing Explainability, 2021-2025
The Collaborative Research Center (TRR 318) „Constructing Explainability“, funded by the DFG at the universities of Bielefeld and Paderborn, addresses the question of how to make algorithmic decisions transparent. Decisions by black-box methods of modern artificial intelligence are especially considered. The central hypothesis is that explanations are most effective if they are co-constructed by explainer and explainee. The mechanisms of co-construction will be investigated by an interdisciplinary consortium to lay the foundations for new paradigms in human-computer interaction to that humans are empowered to make sovereign and informed decisions when interacting with intelligent systems. The TRR is structured in three research areas: A “Explaining”, B “Social practice”, C “Representing and computing explanations”. The research areas consist of interdisciplinary subprojects in which 21 Principal Investigators from Linguistics, Psychology, Media Science, Sociology, Economics and Computer Science from both universities are involved.
Research training group Dataninja, 2021-2025
The goal of trustworthy AI is to offer intelligent methods and agents that produce adaptive behavior in real world scenarios, and are transparent in their decision making as they are able to justify and explain their decisions. The Dataninja research training group is focused on developing novel methods in this area of trustworthy Artificial Intelligence. The goal is to make AI more robust, easier to integrate, more trustworthy, and more secure. Dataninja connects leading research groups in AI in North Rhine-Westphalia and offers young academics an ideal opportunity to establish their own research in the core areas of AI as part of an excellent network.
Competence center Arbeitswelt.Plus - KI in der Arbeitswelt des Industriellen Mittelstandes in OstWestfalenLippe (KIAM), 2020-2025
How will Artificial Intelligence (AI) change the world of work? And how can companies use new technologies? The identification of possible uses for AI and the development of specific solutions pose challenges for small and medium-sized enterprises (SMEs) in particular: a shortage of skilled workers, a lack of technological prerequisites or unclear organizational structures are hurdles that have to be overcome on the way to the introduction of AI-based processes. Above all, there is a lack of work research close to small and medium-sized companies that provides solution and application knowledge about AI in order to relieve companies of the uncertainty before the introduction. These tasks are to be taken over by the Arbeitsswelt.Plus competence center.
RailCampus OWL, 2020-2024
Tomorrow’s mobility is a key factor for the future viability of regions. Transport by rail offers particularly great potential for automation - be it in equipping the wagons, loading and unloading freight trains or running operations. In order to strengthen the Deutsche Bahn in both passenger and freight transport, these opportunities for the application of new technologies must also be used: With the RailCampus OWL, a focal point for these future tasks with high visibility throughout Germany is to be created - as a place of research, development and testing, as well as a campus for studies and further education. The focus of research at the RailCampus OWL will be application-oriented, so that approaches can be tried out directly on site and checked for operability. A group of partners of Campus OWL with diverse scientific resources, DB Systemtechnik with its internationally unique test rigs and test tracks, and DB Cargo AG with internationally recognized know-how in the field of rail technology and operation strengthen the RailCampus OWL. A growing number of medium-sized companies and other partners of the it’s OWL network ensure the transfer to business and society.
Robust Argumentation Machines (RATIO), 2018-2024
The DFG priority program seeks a paradigm shift in which individual facts are replaced by coherent argumentative structures as information units for decision-making, and are systematically and explicitly prepared. For this purpose, new methods are needed that can extract arguments and their relationships from documents as well as new semantic models and ontologies for the deep representation of arguments. New search methods are needed that can index arguments, find relevant arguments for a search query and make them accessible to specific interaction with a human user. In addition, new methods of machine reasoning have to be developed in order to evaluate the implications of arguments and their plausibility. These goals require the cooperation of different disciplines, which are to cooperate for the first time within the framework of a priority program of computer science. Combining competences is the prerequisite for the creation of a paradigm shift, which places argumentative contexts instead of individual facts in the center of information processing and results in a far-reaching potential for new applications in the field of engineering, medicine, finance and online commerce, politics and the humanities, and law.
DataLiteracySkills@OWL (DaLiS@OWL), 2020-2023
In our increasingly data-driven world, the self-determined and secure handling of data is becoming a key competence - not only in science and on the job market, but also in the further development of a digitized society. The corona pandemic is currently making terrifyingly clear how important data skills are for each and every one of us. Because data skills enable us to grasp the importance of parameters such as the R-value, the positive rate or the seven-day incidence and to classify media reports on current, scientific findings on the corona virus. Therefore, the aim of the inter-university joint project DataLiteracySkills @ OWL of the Universities of Bielefeld and Paderborn and the Bielefeld University of Applied Sciences is to strengthen data literacy skills.
it’s OWL: Technology transfer, 2021-2022
The successful technology transfer to medium-sized companies is a unique selling point of it’s OWL. Small and medium-sized companies can use expertise, methods and technologies from the cluster in transfer projects with a university or research institution to solve specific challenges of digital transformation. The projects are easy to apply for and can be implemented quickly. Their effects are directly visible in the company. In this way, medium-sized companies in particular can take important steps on the way to Industry 4.0. The transfer projects make a key contribution to the digitization of processes, products and services. This includes, for example, the intelligent networking and self-optimization of machines and systems, IT security, the design of human-machine interfaces, efficient energy management or new business models.
Product creation is a central task for manufacturing companies: The share of software is constantly increasing and makes the products more and more complex. More than ever before, development requires the combined expertise of different disciplines. AI applications open up far-reaching potential for optimizing processes and increasing the performance of manufacturing companies. Specifically, development capacities can be increased and development times and subsequent manufacturing costs reduced. However, the necessary AI expertise is often missing. Providers of AI solutions often lack access to manufacturing companies. An AI platform should be the solution. The AI Marketplace creates a unique ecosystem that brings together AI experts, vendors and users to tap the full potential of artificial intelligence. The AI Marketplace is a platform that provides a space for secure data exchange and data sovereignty in addition to an intelligent matching of AI service providers and companies. The project consortium of research institutions, networks and companies would like to support especially small and medium-sized enterprises in using artificial intelligence for their product development.
Explainable machine learning for interactive episodic updates of models (EML4U), 2020-2022
Machine Learning (ML) allows complex relationships to be modeled using data. Thus, complex and often grossly simplified mathematical modelling of certain conditions can be avoided. In addition, a new type of functionality opens up: ML models can be adapted to changing requirements and conditions in a data-driven manner. In order to achieve a regular adaptation, data is collected during the use of a model and the model is re-trained taking this information into account. Thus, an episodic update of the ML model takes place.
Bias in AI models, 2020-2022
The term “bias” describes the phenomenon that AI models reflect correlations in data instead of actual causalities when making decisions, even if it is based on non-justifiable and rather historically caused relations. Popular examples are the prediction of the probability of committing a crime based on the ethnicity of a person or the recommendation to employ a person or not based on genders. Since AI models will inherently be more ubiquitous in all fields of society, economy and science in the future, such biases have a large potential impact on marginalized groups and society at large. The project analyzes the impact of data on the learning process of AI models with a focus on language and its influence on different aspects, e.g. building an opinion.
NRW Forschungskolleg: Design of flexible working environments - human-centered use of Cyber-Physical Systems in industry 4.0, 2014-2022
For production companies, the transition to Industry 4.0 opens up great opportunities for modernization and the associated increase in the efficiency of production processes. In addition to the still largely existing technical challenges involved in developing such systems, the role of employees throughout the entire value chain is undergoing considerable change. The seamlessly networked, dynamic and real-time-oriented processes fundamentally change work processes and require more flexible employment. The challenge lies in the development of new social infrastructures, which anticipate rapid technological development and see the human being as the focus of development throughout his entire working life.
ML technologies available in commercial practice cannot meet many practical requirements: data acquisition and analysis are mostly carried out by experts in an offline process. In technical systems, however, a real-time reaction to process changes is required in many cases. Although modern ML technologies such as deep learning have revolutionized areas such as image or speech recognition, these processes are limited by their black box characteristics, the need for large amounts of data to train the process and the often considerable central computing effort. Conversely, the use of ML technologies presents manufacturers of technical systems with new challenges: New product development and engineering approaches are required that leave room for adaptivity. The introduction of new, data-based business models also requires complex adaptation processes. These often overwhelm SMEs in particular, because there are no standardized procedures available for the successful commercialization of learning technical systems. The main goal of the ITS.ML research project is to enable companies, especially SMEs, to utilize the enormous potential of ML along the entire value chain by developing lean ML technologies and bringing them locally into products and production facilities.