Engineering the Neural Collapse Geometry of Supervised-Contrastive Loss (Student Abstract)

摘要

Supervised-contrastive loss (SCL) is an alternative to cross-entropy (CE) for classification tasks that makes use of similarities in the embedding space to allow for richer representations. Previous works have used trainable prototypes to help improve test accuracy of SCL when training under imbalance. In this work, we propose the use of fixed prototypes to help engineering the feature geometry when training with SCL. We gain further insights by considering a limiting scenario where the number of prototypes far outnumber the original batch size. Through this, we establish a connection to CE loss with a fixed classifier and normalized embeddings. We validate our findings by conducting a series of experiments with deep neural networks on benchmark vision datasets.