TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Papers/Self-Supervised Aggregation of Diverse Experts for Test-Ag...

Self-Supervised Aggregation of Diverse Experts for Test-Agnostic Long-Tailed Recognition

Yifan Zhang, Bryan Hooi, Lanqing Hong, Jiashi Feng

2021-07-20Image ClassificationLong-tail LearningTest Agnostic Long-Tailed Learning
PaperPDFCode(official)Code(official)

Abstract

Existing long-tailed recognition methods, aiming to train class-balanced models from long-tailed data, generally assume the models would be evaluated on the uniform test class distribution. However, practical test class distributions often violate this assumption (e.g., being either long-tailed or even inversely long-tailed), which may lead existing methods to fail in real applications. In this paper, we study a more practical yet challenging task, called test-agnostic long-tailed recognition, where the training class distribution is long-tailed while the test class distribution is agnostic and not necessarily uniform. In addition to the issue of class imbalance, this task poses another challenge: the class distribution shift between the training and test data is unknown. To tackle this task, we propose a novel approach, called Self-supervised Aggregation of Diverse Experts, which consists of two strategies: (i) a new skill-diverse expert learning strategy that trains multiple experts from a single and stationary long-tailed dataset to separately handle different class distributions; (ii) a novel test-time expert aggregation strategy that leverages self-supervision to aggregate the learned multiple experts for handling unknown test class distributions. We theoretically show that our self-supervised strategy has a provable ability to simulate test-agnostic class distributions. Promising empirical results demonstrate the effectiveness of our method on both vanilla and test-agnostic long-tailed recognition. Code is available at \url{https://github.com/Vanint/SADE-AgnosticLT}.

Results

TaskDatasetMetricValueModel
Image ClassificationPlaces-LTTop 1 Accuracy40.9TADE
Image ClassificationPlaces-LTTop-1 Accuracy41.3TADE
Image ClassificationCIFAR-10-LT (ρ=10)Error Rate9.2TADE
Image ClassificationCIFAR-10-LT (ρ=10)Error Rate10.3RIDE
Image ClassificationCIFAR-100-LT (ρ=50)Error Rate46.1TADE
Image ClassificationCIFAR-100-LT (ρ=10)Error Rate36.4TADE
Image ClassificationImageNet-LTTop-1 Accuracy61.4TADE(ResNeXt101-32x4d)
Image ClassificationImageNet-LTTop-1 Accuracy58.8TADE(ResNeXt-50)
Image ClassificationCIFAR-100-LT (ρ=100)Error Rate50.2TADE
Image ClassificationCIFAR-10-LT (ρ=100)Error Rate16.2TADE
Few-Shot Image ClassificationPlaces-LTTop 1 Accuracy40.9TADE
Few-Shot Image ClassificationPlaces-LTTop-1 Accuracy41.3TADE
Few-Shot Image ClassificationCIFAR-10-LT (ρ=10)Error Rate9.2TADE
Few-Shot Image ClassificationCIFAR-10-LT (ρ=10)Error Rate10.3RIDE
Few-Shot Image ClassificationCIFAR-100-LT (ρ=50)Error Rate46.1TADE
Few-Shot Image ClassificationCIFAR-100-LT (ρ=10)Error Rate36.4TADE
Few-Shot Image ClassificationImageNet-LTTop-1 Accuracy61.4TADE(ResNeXt101-32x4d)
Few-Shot Image ClassificationImageNet-LTTop-1 Accuracy58.8TADE(ResNeXt-50)
Few-Shot Image ClassificationCIFAR-100-LT (ρ=100)Error Rate50.2TADE
Few-Shot Image ClassificationCIFAR-10-LT (ρ=100)Error Rate16.2TADE
Generalized Few-Shot ClassificationPlaces-LTTop 1 Accuracy40.9TADE
Generalized Few-Shot ClassificationPlaces-LTTop-1 Accuracy41.3TADE
Generalized Few-Shot ClassificationCIFAR-10-LT (ρ=10)Error Rate9.2TADE
Generalized Few-Shot ClassificationCIFAR-10-LT (ρ=10)Error Rate10.3RIDE
Generalized Few-Shot ClassificationCIFAR-100-LT (ρ=50)Error Rate46.1TADE
Generalized Few-Shot ClassificationCIFAR-100-LT (ρ=10)Error Rate36.4TADE
Generalized Few-Shot ClassificationImageNet-LTTop-1 Accuracy61.4TADE(ResNeXt101-32x4d)
Generalized Few-Shot ClassificationImageNet-LTTop-1 Accuracy58.8TADE(ResNeXt-50)
Generalized Few-Shot ClassificationCIFAR-100-LT (ρ=100)Error Rate50.2TADE
Generalized Few-Shot ClassificationCIFAR-10-LT (ρ=100)Error Rate16.2TADE
Long-tail LearningPlaces-LTTop 1 Accuracy40.9TADE
Long-tail LearningPlaces-LTTop-1 Accuracy41.3TADE
Long-tail LearningCIFAR-10-LT (ρ=10)Error Rate9.2TADE
Long-tail LearningCIFAR-10-LT (ρ=10)Error Rate10.3RIDE
Long-tail LearningCIFAR-100-LT (ρ=50)Error Rate46.1TADE
Long-tail LearningCIFAR-100-LT (ρ=10)Error Rate36.4TADE
Long-tail LearningImageNet-LTTop-1 Accuracy61.4TADE(ResNeXt101-32x4d)
Long-tail LearningImageNet-LTTop-1 Accuracy58.8TADE(ResNeXt-50)
Long-tail LearningCIFAR-100-LT (ρ=100)Error Rate50.2TADE
Long-tail LearningCIFAR-10-LT (ρ=100)Error Rate16.2TADE
Generalized Few-Shot LearningPlaces-LTTop 1 Accuracy40.9TADE
Generalized Few-Shot LearningPlaces-LTTop-1 Accuracy41.3TADE
Generalized Few-Shot LearningCIFAR-10-LT (ρ=10)Error Rate9.2TADE
Generalized Few-Shot LearningCIFAR-10-LT (ρ=10)Error Rate10.3RIDE
Generalized Few-Shot LearningCIFAR-100-LT (ρ=50)Error Rate46.1TADE
Generalized Few-Shot LearningCIFAR-100-LT (ρ=10)Error Rate36.4TADE
Generalized Few-Shot LearningImageNet-LTTop-1 Accuracy61.4TADE(ResNeXt101-32x4d)
Generalized Few-Shot LearningImageNet-LTTop-1 Accuracy58.8TADE(ResNeXt-50)
Generalized Few-Shot LearningCIFAR-100-LT (ρ=100)Error Rate50.2TADE
Generalized Few-Shot LearningCIFAR-10-LT (ρ=100)Error Rate16.2TADE

Related Papers

Automatic Classification and Segmentation of Tunnel Cracks Based on Deep Learning and Visual Explanations2025-07-18Adversarial attacks to image classification systems using evolutionary algorithms2025-07-17Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy2025-07-17Federated Learning for Commercial Image Sources2025-07-17MUPAX: Multidimensional Problem Agnostic eXplainable AI2025-07-17Hashed Watermark as a Filter: Defeating Forging and Overwriting Attacks in Weight-based Neural Network Watermarking2025-07-15Transferring Styles for Reduced Texture Bias and Improved Robustness in Semantic Segmentation Networks2025-07-14FedGSCA: Medical Federated Learning with Global Sample Selector and Client Adaptive Adjuster under Label Noise2025-07-13