Dynamic collaborative learning with heterogeneous knowledge transfer for long-tailed visual recognition

Hao Zhou, Tingjin Luo, Yongming He

Abstract: Highlights•We introduce a multi-experts collaboration model to collaboratively learn the classifier for the head, medium, and tail samples.•We design a dynamic adaptive sinusoidal weight to adjust the model’s attention to different experts in training.•Heterogeneous knowledge transfer learning is proposed to enhance message passing between experts with different specialties.