In this work, we introduce ProtoNCD, a novel approach to novel class discovery (NCD) that leverages prototypical parts for enhanced interpretability. ProtoNCD extends the ProtoPool methodology to the NCD setting, employing techniques such as knowledge distillation and specialized prototypical parts initialization. Through comprehensive experiments on the CUB-200-2011 dataset, we demonstrate the efficacy of ProtoNCD and its pivotal role in explaining how the reasoning of known classes influences predictions for those newly discovered.

• Tomasz Michalski, Jagiellonian University, Doctoral School of Exact and Natural Sciences & Faculty of Mathematics and Computer Science
• Dawid Rymarczyk, Jagiellonian University, Faculty of Mathematics and Computer Science
• Daniel Barczyk, Jagiellonian University, Faculty of Mathematics and Computer Science
• Bartosz Zieliński, Jagiellonian University, Faculty of Mathematics and Computer Science & IDEAS NCBR

• pytorch
• wandb
• sklearn
• pandas
• numpy
• tqdm

python ProtoNCD.py --c scripts/discover_freeze_pretrained_slots.yaml

This work was funded by the National Science Centre (Poland) grant no. 2022/47/B/ST6/03397. We gratefully acknowledge Polish high-performance computing infrastructure PLGrid (HPC Centers: ACK Cyfronet AGH) for providing computer facilities and support within computational grant no. PLG/2023/016555.