PhD position H/F - Assessing and Explaining Deep Learning Model Compression at CESI

This is a joint French-German PhD thesis on Green AI and explainable AI for automatic speech recognition, with research on self-supervised model compression, quantization, pruning, and model interpretability. The thesis will be registered in ENSAM (France) and THM (Germany), with 18 months in France at CESI LINEACT’s Campus de Nancy and 18 months in Germany at KITE in Friedberg, followed by a 3-year duration and a starting date as soon as possible. The work requires strong mathematical, programming, and AI skills, with PyTorch or TensorFlow experience appreciated, and will be co-supervised by Leila Ben Letaifa, Yuehua Ding, Michael Guckert, and Christin Seifert.