luhai_slides_template.pptx
PPTX,
1 MB
Individual thesis topics
We are looking for students who would like to work on our future research projects. Depending on your skills and future plans, we will work together to find a topic in our research areas that best suits you. In your thesis, you will come into contact with emerging topics with a focus on AI. The following topics are examples of what such work might look like.
Topic Areas
-
General AutoML
Automated Machine Learning (AutoML) streamlines the development of ML pipelines and supports practitioners in critical design decisions that are crucial for optimized performance. Key technical pillars driving modern AutoML research include:
(I) sample-efficient Bayesian Optimization
(II) Multi-Fidelity optimization for expensive, large-scale models
(III) Neural Architecture Search (NAS) for designing neural networks tailored to best performance
As a prerequisite, you should
- be experienced in coding in Python,
- have taken at least one Machine Learning course and also the AutoML lecture or the corresponding MOOC by Prof. Lindauer and colleagues.
-
Human-Centered AutoML
We are offering thesis topics in the area of Human-Centered AutoML - an area that combines the automation of machine learning pipelines with the interpretability and usability that real-world users demand. Our thesis topics are ideal for students who want to work at the intersection of core AutoML research, explainability of AutoML methods, and the utilisation of foundation models (incl. LLMs) for AutoML.
The required prerequisites may vary greatly depending on the concrete topic. As a general guideline, we might expect:
- Knowledge in (Automated) Machine Learning and Deep Learning, demonstrated by attending relevant courses, such as Machine Learning, Automated Machine Learning, Interpretable Machine Learning, and/or Deep Learning.
- Practical experience with Python, numpy, pandas, torch, and related libraries4
Contact: Marcel Wever or Lukas Fehring
-
Green AutoML
The rapid growth of AI technologies has led to increasing concerns about their environmental impact, particularly due to high energy consumption during model development, training, and inference. Green AutoML addresses these challenges by focusing on three main areas: making Machine Learning greener through AutoML, improving the efficiency of AutoML itself, and applying AutoML to sustainability-related problems. Thesis topics will explore specific challenges within one of these areas, contributing to the development of more sustainable and responsible AI solutions. As a prerequisite, you should
- be experienced in coding in Python,
- have taken at least one Machine Learning-related course, preferably by Prof. Lindauer.
Contact: Daphne Theodorakopoulos and Leona Hennig
-
Reinforcement Learning
Our focus is on making Reinforcement Learning (RL) applicable to real-world problems. We do this using contextual RL, exploring the role of representations and configuring hyperparameters, architectures, and algorithms. Thesis experiments will usually be conducted in simulation environments (with optional deployment to robots). They will generally focus on either improving the generalization or efficiency of existing algorithms or studying specific aspects of these algorithms. We recommend the following skills:
- Experience with Deep Learning in Python (either PyTorch or JAX)
- The RL lecture by Prof. Lindauer (or equivalent RL experience)
- Depending on the topic, additional knowledge of Deep Learning, HPO, or NAS
Contact: Theresa Eimer and Aditya Mohan
Additional Topics with External Partners
-
ML for Solar Energy Research - In Cooperation with the ISFH [MSc]
A new generation of solar cells is based on the material perovskite. At the Institute for Solar Energy Research in Hamelin (ISFH) , intensive research is currently being conducted to enable the production of perovskite solar cells through the evaporation of various starting materials. The processes are not yet sufficiently reproducible, meaning that identical process parameter settings lead to different degrees of efficiency. As part of your master's thesis, you will support the research group in understanding and stabilizing the manufacturing process. To do this, you will use state of the art AI-methods to analyze the existing process data in the form of time series and examine it for patterns and correlations, e.g., correlations between individual process parameters and efficiency. Maybe XAI methods could be used here as well. If these correlations can be identified, the next step would be to optimize the process parameters using a Machine Learning model, directly contributing to more efficient solar cells and sustainable energy production. The thesis will be done at the Institute of AI. Regular visits to the ISFH laboratory will help to understand the experimental challenges, process data collection, and structure, and to evaluate feedback from the analysis on the experiment.
Contact: Daphne Theodorakopoulos
Interested?
The exact procedure of a thesis, together with a rough idea of what we expect from theses, is described here.
It is important to us that the appropriate background knowledge is available so that a thesis has a chance of a positive conclusion. In order to be able to assess this accordingly, we would ask you to send us the following points:
Proposed topic or topic area(s)
What previous knowledge is available? What ML-related courses have been taken for this?
- A self-assessment from -- to ++ on the following topics:
- Coding in Python
- Coding with PyTorch
- Ability to implement a Deep Learning paper
- Ability to implement a reinforcement learning paper
- Ability to understand and execute a foreign codebase
If you are generally interested in writing a thesis with us but have not decided on any of the above topics, please email m.lindauer@ai.uni-hannover.de with the above information.
If you are interested in a specific topic indicated above, please send an email directly to the contact person indicated in the topic. The email addresses can be found on the personal pages.
