The paper discusses the role of hyperparameters in aggregation models in digital pathology. The authors highlight the potential bias in evaluation methods due to fixed aggregation model hyperparameters. They argue that there’s a co-dependency between feature extractor models and aggregation model hyperparameters, which can skew performance comparability. The paper also emphasizes the need for a comprehensive approach to understand the relationship between feature extractors and aggregation models, to provide a more accurate assessment of feature extractor models in digital pathology.
Publication date: 29 Nov 2023
Project Page: https://arxiv.org/abs/2311.17804
Paper: https://arxiv.org/pdf/2311.17804
