COSEAL - Configuration and Selection of Algorithms

The COSEAL research group is an international consortium of researchers from all over the world (e.g. Belgium, Canada, Ireland, Denmark and Germany) which addresses current challenges from Algorithm Selection, Algorithm Configuration and Machine Learning.

DemoRESILdigital: Democratic resilience in times of online-propaganda, fake news, fear- and hate speech

Reducing the moderation effort of user comments with the help of automation using text analytical methods (MODERAT!)

In recent years, a rapid increase in racist, political and religiously motivated hate commentary has led many newspaper editors to deactivate their online comment functions on their websites. While this is understandable from an economic point of view for the individual publishers, serious problems for the public discourse arise in view of restriction quotas of up to 50%. The MODERAT! project aims to use an integrative and interdisciplinary approach to develop software tools and a web platform that will enable operators to moderate web debates with significantly less effort. Comments are analyzed automatically, so that only a small number of critical comments have to be viewed manually. In this way, media houses and publishers should be able to offer web debates again on their own websites and thus enter into a more active exchange with the readership.

PPP Australien 2020: Detection and classification of malicious virtual grassroots influence campaigns

Joint research project: Identifying, verifying, and stopping concealed propaganda attacks via online media - Sub-Project: Coordination of simulation, identification, and defense against concealed propaganda attacks

This project addresses the identification and verification of (semi-)automated concealed propaganda in online media (e.g. in social networks) as an interdisciplinary approach.

Benchmarking Network

The Benchmarking Network is an initiative that has emerged in summer 2019, with the idea to consolidate and to stimulate activities on benchmarking iterative optimization heuristics such as local search algorithms, swarm intelligence techniques, model- and/or surrogate-based heuristics, etc - in short, all algorithms that work by a sequential evaluation of solution candidates.

Architectures for Evolutionary Multi-Objective Algorithms to support User Expertise Integration

This project strives to analyze existing algorithm architectures and design new structures such that integration of expertise becomes seamlessly possible. Users should be free of the need to redesign or deeply understand the general algorithms structure.

Using Evolutionary Algorithms for Diversity Optimization

This project is one of the STSMs (Short Term Scientific Missions) of the COST (European Cooperation in Science and Technology) Action CA15140 on "Improving Applicability of Nature-Inspired Optimisation by Joining Theory and Practice (ImAppNIO)", which aims at bridging the gap between theory and practice of nature inspired optimization algorithms.

ERCIS Omni-Channel Lab - powered by Arvato

The ERCIS Omni-Channel Lab, is a research project in collaboration with Arvato CRM Solutions. The Lab combines the academic research and teaching activities of ERCIS and Arvato’s practical experience of delivering tech-enabled Omni-Channel CRM solutions from 110 global locations for many of the world’s best-known brands. Its areas of investigation focus around ‘Processes’, ‘Data’ and ‘Analytics’.

Instance-Based Algorithm Selection of Inexact TSP solvers

The Travelling Salesperson Problem (TSP) is arguably the most prominent NP-hard combinatorial optimisation problem. Given a set of n cities and pairwise distances between those, the objective in the TSP is to find the shortest round-trip or tour through all cities, i.e., a sequence in which every city is visited exactly once, the start and end cities are identical, and the total length of the tour is minimal. The Euclidean TSP has important applications, e.g., in the fabrication of printed circuit boards as well as in transportation and logistics. We aim at constructing an instance-based algorithm selection model in order to improve the current state-of-the-art solver.

Validation of automated identification of Social Bots

9th International Conference on Evolutionary Multi- Criterion Optimization, Münster 19. - 22.03.2017

EMO 2017 is the 9th International Conference on Evolutionary Multi- Criterion Optimization, aiming to continue the success of previous EMO conferences. We will bring together both the EMO and the multiple criteria decision making (MCDM) communities and moreover focus on solving real-world problems in government, business and industry. The classical EMO format will be supplemented by an EMO competition.

Hybridization of indicator-based metaheuristics with modern local search methods in multiobjective optimization

The project realizes international expertise exchange between German and Mexican researchers in the context of hybrid evolutionary multi-objective optimization. A special focus lies on integrating local search into state-of-the-art meta-heuristics like SMS-EMOA and dP-EMOA.

Google Summer of Code 2015: Improving mlr's hyperparameter and tuning system for efficient model selection

This "Google Summer of Code 2015" project aims at enriching mlr's hyperparamter system with a-priori knowledge. The goal of the project is to make the hyperparameter configuration and tuning more flexible, efficient and convenient.

Addressing Current Challenges in Evolutionary Multi-Objective Optimization: Indicator-based Selection, Convergence and Applicability

This project aims to initiate and intensify bi-lateral collaboration of researchers from Brazil and Germany under the umbrella of current research topics in the domain of evolutionary multi-objective optimization.

Structural optimization of distributed energy supply systems - applied to chemical parks

The project aimed at setting up a method for structural optimization that enables the planning engineer to design a distributed energy supply system of near-optimal quality.