Hi, in principle yes. But the problem with GANs can be the compute time, depending how it is done. It might be feasible with some pre-trained models ? So there is the danger of not getting own results in due time. But if you can explain what you did, and what the problems were if you don't get own results, this is fine as a contribution for the final project. Regards, Jochen Garcke On 01/07/2022 08:57, Moritz Christopher Kappes wrote:
Dear team of the practical Lab, my excercise partner and I (Georg S. & Moritz K.) wanted to do the following kaggle competition as a final project: https://www.kaggle.com/competitions/gan-getting-started/overview
We think it is a good extension of the deep learning topic, as we can explain a bit what GANs are and present another architecture of a Deep Neural Network. We also think that it would maybe be interesting to see some of the generated images in the presentation.
Do you think this topic is suitable as a final project?
Kind regards, Moritz Kappes _______________________________________________ mllab-22 mailing list mllab-22@ins.uni-bonn.de https://mailman.ins.uni-bonn.de/mailman/listinfo/mllab-22
-- Prof. Dr. Jochen Garcke garcke@ins.uni-bonn.de jochen.garcke@scai.fraunhofer.de phone: +49-228 73 69838 phone: +49-2241 14 4015 http://garcke.ins.uni-bonn.de http://scai.fraunhofer.de/ndv Universität Bonn Fraunhofer-Institut SCAI Institute for Numerical Simulation Numerical Data-Driven Prediction Friedrich-Hirzebruch-Allee 7 Schloss Birlinghoven 53115 Bonn, Germany 53757 Sankt Augustin, Germany