Nested Collaborative Learning for Long-Tailed Visual Recognition

CVPR 2022  ·  Jun Li, Zichang Tan, Jun Wan, Zhen Lei, Guodong Guo ·

The networks trained on the long-tailed dataset vary remarkably, despite the same training settings, which shows the great uncertainty in long-tailed learning. To alleviate the uncertainty, we propose a Nested Collaborative Learning (NCL), which tackles the problem by collaboratively learning multiple experts together. NCL consists of two core components, namely Nested Individual Learning (NIL) and Nested Balanced Online Distillation (NBOD), which focus on the individual supervised learning for each single expert and the knowledge transferring among multiple experts, respectively. To learn representations more thoroughly, both NIL and NBOD are formulated in a nested way, in which the learning is conducted on not just all categories from a full perspective but some hard categories from a partial perspective. Regarding the learning in the partial perspective, we specifically select the negative categories with high predicted scores as the hard categories by using a proposed Hard Category Mining (HCM). In the NCL, the learning from two perspectives is nested, highly related and complementary, and helps the network to capture not only global and robust features but also meticulous distinguishing ability. Moreover, self-supervision is further utilized for feature enhancement. Extensive experiments manifest the superiority of our method with outperforming the state-of-the-art whether by using a single model or an ensemble.

PDF Abstract CVPR 2022 PDF CVPR 2022 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Long-tail Learning CIFAR-100-LT (ρ=100) NCL(ResNet32) Error Rate 46.7 # 17
Long-tail Learning CIFAR-100-LT (ρ=50) NCL(ResNet32) Error Rate 43.2 # 14
Long-tail Learning CIFAR-10-LT (ρ=100) NCL(ResNet32) Error Rate 15.3 # 9
Long-tail Learning CIFAR-10-LT (ρ=50) NCL(ResNet32) Error Rate 13.2 # 5
Long-tail Learning ImageNet-LT NCL(ResNeXt-50) Top-1 Accuracy 58.4 # 17
Long-tail Learning ImageNet-LT NCL(ResNet-50) Top-1 Accuracy 57.4 # 22
Long-tail Learning iNaturalist 2018 NCL(ResNet-50) Top-1 Accuracy 74.2% # 15
Long-tail Learning Places-LT NCL(ResNet-152) Top-1 Accuracy 41.5 # 12

Methods