Set2Seq Transformer: Temporal and Position-Aware Set Representations for Sequential Multiple-Instance Learning
arXiv:2408.03404v3 Announce Type: replace-cross
Abstract: In many real-world applications, modeling both the internal structure of sets and their temporal relationships is essential for capturing complex underlying patterns. Sequential multiple-instan…