Hello there 👋

I'm Tim Wibiral

Data Scientist | Software Developer | AI Enthusiast

About me

Passionate about excellent software, interpretable machine learning and the future of AI. Especially interested in the foundational mechanisms behind neural networks and the ethical implications of AI.

I recently finished my master's thesis on LossVal, an efficient data valuation method for neural networks, in collaboration with Porsche AG.

Contact

Mail

What I am reading

Publications

LossVal: Efficient Data Valuation for Neural Networks
arxiv preprint (2024)
Weighted Pairwise Difference Learning*
LWDA’24: Lernen, Wissen, Daten, Analysen. September 23–25, 2024, Würzburg, Germany (2024)
Examining the Role of Genuine Emotions for Trustworthy AI
HAI '23: Proceedings of the 11th International Conference on Human-Agent Interaction (2023)
JavaCHR – A Modern CHR-Embedding in Java
Bachelor's Thesis (2022)

*Co-author

Software Projects

LossVal

An efficient data valuation approach for neural networks. GitHub: github.com/twibiral/LossVal

Obsidian Execute Code Plugin

Plugin for Obsidian.md that allows the user to execute code in a note. GitHub: github.com/twibiral/obsidian-execute-code

ObsiDOOM

Plugin for Obsidian.md to play retro games like DOOM or Prince of Persia. GitHub: github.com/twibiral/ObsiDOOM

JavaCHR

An object-oriented embedding of Constraint Handling Rules (CHR) in Java. GitHub: github.com/twibiral/JavaCHR

Showroom for Computer Vision Filters

Website that shows the effects of different images noises and filters when they are combined. Link to the website: tim-wibiral.de/ComputerVisionShowroom GitHub: github.com/twibiral/ComputerVisionShowroom

Deep-Q-Ensembles and -Soups for Atari Games

Implementation of Deep-Q-Learning for Atari games combined with Ensembles and Soups of multiple independent models. Project work for the "Deep Reinforcement Learning" project course. GitHub: github.com/twibiral/DQNwithEnsemblesAndSoups