DACBench is a benchmarking library that allows for the dynamic control of hyperparameters of different target algorithms (e.g. EAs or neural nets). It has recently been published at IJCAI and continues to be extended and improved through collaborations on new features and benchmarks. As a HiWi for DACBench, your task will be to work on maintenance as well as technical features and quality of life improvements (e.g. speed up logging, make containers for dependencies, parallelize benchmarks). If interested, you can also join efforts for more research-focused projects. In any case you're free to divide your time as you see fit, but will be expected to join hackathons and meetings with researchers from different universities and backgrounds.
Fluency in English (speaking & writing) as well as excellent knowlege of python. Knowledge of RL is a bonus, but not required. As we have several features planned, knowledge of the following items can be a plus and should be mentioned: Webdesign, interfacing python and C++ code and parallelization.
Contact person: Theresa Eimer