Computational Science Hub - CSH

Projects

SimLearn

Status: in progress

Description

Machine learning methods based on existing training data have proven to be very effective in identifying patterns and implicit dependencies in complex situations with many parameters and in providing classification, prediction and decision support with the models learned. In practice, however, the large amounts of correctly labeled training data required for such approaches are often not available.

Based on actual application examples from the agricultural sector, SimLearn examines the suitability of a new approach in which existing operative knowledge codified in simulation models is combined iteratively with the increasing insights of learned models: Extensive synthetic training data sets are generated by existing simulation models. A learning system initiated on such data will then be extended and improved by empirical data collected of actual farms. This combination fills gaps in the existing database and enables improved training. The result is a learned, more powerful model of the observed reality with improved usage potentials.

People
Departments
Fruther Informationen

Considering the minimum distance in model-based order processing

Status: in progress

Description

Order picking describes the process of retrieving inventory items from their storage locations to satisfy customer orders. It has long been identifed as the most laborious and costly warehouse operation. Research papers on order picking have mostly concentrated on picker routing problems which aim at determining a cost-minimal order picking tour along the storage locations defined by a customer order. To reduce the risk of infection in warehouses in times of COVID-19 (or other infectious diseases), legal requirements (such as keeping a minimum distance) must be considered when determining the routes through the warehouse. The goal of the research project is to develop an exact algorithm that determines the routes through a warehouse such that the distances to be covered by the order pickers are minimized and the required minimum distances are met. The algorithm shall be realized as a software module that can be integrated into existing systems for decision support.

People
Departments

More than meets the eyes: Machine Learning in and for agent-based modeling

Status: in progress

Description

In recent years, many scholars praised the seemingly endless possibilities of using machine learning (ML) techniques in and for agent-based simulation models (ABM). To get a more comprehensive understanding of the opportunities, we conduct a systematic literature review (SLR) and classify the literature on the application of ML in and for ABM according to a theoretically derived classification scheme. We do so to investigate how exactly machine learning has been utilized in agent-based models in different disciplines so far and to identify the most important use cases in the literature. We find that, indeed, there is a broad range of possible applications of ML that might help ABMs to unfold their full potential. Further, we see that, ML is so far mainly used in ABM for the modeling of adaptive agents equipped with experience learning. While these are the most frequent, there is also a variety of many more interesting applications which do not directly meet the eye. This being the case, researchers should dive deeper into the analysis of when and how which kinds of ML techniques can support ABM, e.g. by conducting a more in-depth analysis and comparison of different use cases. Nonetheless, as the application of ML in and for ABM comes at certain costs, researcher should not use ML for ABMs just for the sake of doing it.

People
Departments

Forecasting Baden-Württemberg's GDP Growth: MIDAS Regressions versus Dynamic Mixed-Frequency Factor Models

Status: in progress

Description

Germany's economic composition is heterogenous across regions which makes regional economic projections based on German GDP growth unreliable. In this paper, we develop forecasting models for Baden-Württemberg's economic growth, a regional economy that is dominated by small and medium-sized enterprises with a strong focus on foreign trade. For this purpose, we evaluate the backcasting and nowcasting performance of MIDAS regressions with forecast combinations against an approximate dynamic mixed-frequency factor model. Considering a wide range of regional, national, and global predictors, we find that our high-dimensional models outperform benchmark time series models. Surprisingly, we also find that combined forecasts based on simple single-predictor MIDAS regressions are able to outperform more sophisticated dynamic factor models.

People
Departments

COVID-19 Simulation for Baden-Württemberg

Status: in progress

Description

Mathematical epidemiology has evolved over the last decades. In combination with modern simulation techniques it now provides powerful methods to support decision making, also in the current context of the Corona-crisis. The aim of this project is to develop a dynamic spatial model for Baden-Württemberg based on reaction-diffusion equations which allows to investigate the spread of infections under various scenarios and counter measures. The purpose of this project, therefore, is to study how mathematical models can help prevent a second wave of infections which potentially would overwhelm the health system.

People
Departments
Further information and media response

The Living Heart Project

Description

The “Living Heart” project combines the competencies from mathematics, cardiovascular research, cardiology and medical engineering as well as expertise from regulators with the aim to develop realistic simulation models of the human heart, based on mathematical models and computer aided engineering (CAE).

The main objective is to enhance the understanding of the heart, and thus to contribute to both, the development and testing of new medical products (e.g. pacemakers). Furthermore, new approaches are developed which allow to transpose a 3D-scan of a patient's heart into a model to improve not only diagnostics but also allow for specific adjustments in a patient's individual the therapy.

People
Further information and media response

 

 

Research Director

Prof. Dr. Thilo Streck
(E-Mail)