Invert color theme

Research at the MALEO Group

The world is multi-objective ...

The MALEO group was established in October 2023 at the University of Paderborn. We have diverse research foci within the general research domain of Trustworthy Artificial Intelligence and extensive national and international collaborations.

Our research addresses real-world problems characterised by various complexities that pose challenges for traditional optimisation algorithms. These include stochasticity, noise, and dynamism, and many problems exhibit a black-box nature, where the objective function is unknown, and information is gathered solely through evaluation. To address these challenges, we employ Randomised Search Heuristics, with a particular emphasis on nature-inspired Evolutionary Computation methods. Our goal is to approximate solutions and comprehensively understand the strengths and weaknesses of these algorithms, conducting both empirical assessments and theoretical analyses through time complexity analysis.

One of the hardest challenges are problems involving the systematic and simultaneous optimization of multiple conflicting objectives. Recognizing the inherent difficulty of solving Multi-Objective Optimization problems exactly, we leverage optimization techniques from evolutionary computation to approximate optimal compromises. Our special focus is on multi-modality, adding depth to our exploration of these multifaceted challenges.

In the realm of Algorithm Benchmarking, our group assesses the performance of nature-inspired techniques and contributes to algorithm design from empirical and theoretical perspectives. Automated Algorithm Selection and Configuration, key facets of our work, involve the automated identification and/or configuration of the best-suited algorithm for a given problem. For this, we e.g. employ Exploratory Landscape Analysis (ELA), where problem properties are matched to known algorithms' performance. Artificial Intelligence and Machine Learning techniques, particularly Deep Learning and Classification approaches, play a pivotal role in constructing accurate and efficient selection models. Collaborating closely with the Configuration and Selection of Algorithms (COSEAL) research group, our team focuses on vehicle routing and continuous black-box optimization. Furthermore, our group delves into the realms of Data Stream Mining and Computational Social Sciences. In collaboration with the University of Münster, and within the Social Media Competence Center of the European Research Center for Information Systems (ERCIS) we actively engage in the detection of propaganda and disinformation campaigns in online media, showcasing our commitment to cutting-edge research in these emerging fields.

For even more in-depth information scan our publications.

  • Multi-Objective Optimisation
  • Randomised Search Heuristics (RSH), in particular Evolutionary Algorithms
  • Artificial Intelligence (AI) and Machine Learning
  • Automated AI (AutoAI), in particular Automated Algorithm-Selection and -Configuration
  • Data Stream Mining
  • Data Science
  • Computational Social Science

Ongoing Projects and Collaboration

Cost Action CA22137 - Randomised Optimisation Algorithms Research Network (ROAR-NET)

This COST Action aims at making randomised optimisation algorithms widely competitive in practice by identifying and reducing obstacles to their adoption at the scientific, technical, economic, and human levels. It focuses on meeting the needs of practitioners, from whose activities the economic value of optimisation solvers stems. These needs are taken as the driving force for new theoretical, methodological, and technical advances leading to the sustainable development of widely available software tools, training materials and programmes, and ultimately to more extensive acceptance and deployment of these methods.

Read more

HybriD - Real-Time Detection and Identification of Hybrid Disinformation Campaigns in Online Media

The goal of the BMBF-funded joint project “Real-Time Detection and Identification of Hybrid Disinformation Campaigns in Online Media (HybriD)” is to develop a software-based analysis tool that helps experts to better assess disinformation campaigns. The interdisciplinary team of computer scientists, communication scientists and IT security service providers combines machine learning with human expertise to identify disinformation campaigns. The analysis tool will make it possible to evaluate large volumes of data from online media and social networks in real time and to detect temporal patterns. Using this database, experts can comprehensively assess disinformation campaigns and their impact. These findings are then returned to the tool, allowing new patterns to be learned. Iteratively, general statements about disinformation patterns can be made and so-called “archetypes” of disinformation can be identified. This is intended to answer the question given at the outset about the emergence and spread of disinformation as well as the possibilities for action.

Read more

Transregional Collaborative Research Centre TRR 318 "Constructing Explainability"

Often, technical explanations presuppose knowledge about AI and are therefore difficult to comprehend. In the Transregional Collaborative Research Centre 318 “Constructing Explainability”, researchers are developing ways to involve users in the explanation process and thus create co-constructive explanations. Therefore, the interdisciplinary research team investigates the principles, mechanisms, and social practices of explaining and how these can be taken into account in the design of AI systems. The goal of the project is to make explanatory processes comprehensible and to create understandable assistance systems.

Read more