To apply,
please send a copy of your CV and a cover letter outlining your research
interests and experience to [cihan.ates@kit.edu]. Please
also include any relevant publications, presentations, or awards. We review
applications on a rolling basis and will contact qualified candidates for an
interview. We look forward to receiving your application and potentially
welcoming you to our team!
Improving Product Valuation in the Energy Market via Generative Adversarial Networks (GANs)
External Master Thesis
Financial
product valuation is essential to support trading and risk management
decisions. A standard method to perform the valuation of financial
products under uncertainty is to estimate their return for alternating
scenarios. These scenarios are usually simulated with Monte-Carlo
methods. For
sophisticated products, this leads to a considerable number of
simulations and high computational expenses. The goal of this study is to utilize the power of GANs to
generate realistic market scenarios to depict the uncertainties in
market conditions.
The
project will be conducted as an external master thesis in collaboration
with Energie Baden-Wรผrttemberg (EnBW) and the student will be employed
at EnBW during the work. The implementation has a significant potential
and we aim to publish the outcomes in international scientific journals.
Requirements:
Modeling of Fuel Injection in Aero-engines
with Causal Generative Machine Learning Models
Master Thesis
One of the advancing frontiers of machine learning is generative modelling.
Such models have a probabilistic nature and rely on the discovered patterns in
the data, which can later be used to create new examples indistinguishable from the original dataset.
In this project, our goal
is to generate a model which can mimic the statistics of fuel droplets in aero-engine
combustors. The training data will be provided from highly resolved reactive CFD
simulations. The student will test alternative generative ML models to
predict the droplet sizes and trajectories along the combustor and evaluate the
predictive accuracy of the models at various operating conditions. Another goal is to test different feedback policies to imrove the causality in model predictions.
Requirements:
created with
Website Builder Software .