Unequal-Training for Deep Face Recognition With Long-Tailed Noisy Data

By Yaoyao Zhong, Weihong Deng, Mei Wang, Jiani Hu, Jianteng Peng, Xunqiang Tao, Yaohai Huang. 2019

129 citations recorded by the source corpus.

Large-scale face datasets usually exhibit a massive number of classes, a long-tailed distribution, and severe label noise, which undoubtedly aggravate the difficulty of training. In this paper, we propose a training strategy that treats the head data and the tail data in an unequal way, accompanying with noise-robust loss functions, to take full advantage of their respective characteristics. Specifically, the unequal-training framework provides two training data streams: the first stream applies the head data to learn discriminative face representation supervised by Noise Resistance loss; the second stream applies the tail data to learn auxiliary information by gradually mining the stable discriminative information from confusing tail classes. Consequently, both training streams offer complementary information to deep feature learning. Extensive experiments have demonstrated the effectiveness of the new unequal-training framework and loss functions. Better yet, our method could save a significant amount of GPU memory. With our method, we achieve the best result on MegaFace Challenge 2 (MF2) given a large-scale noisy training data set.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Signal Processing

Citations in the corpus

Long-Tailed Class Incremental Learning (2022, background): “However, data sampled from the real world often follows a long-tailed distribution [14,24,40,49] in which some classes have many more samples than others.”
Decoupling Representation and Classifier for Long-Tailed Recognition (2019, background): “Long-tailed recognition has attracted increasing attention due to the prevalence of imbalanced data in real-world applications (Wang et al., 2017; Zhou et al., 2017; Mahajan et al., 2018; Zhong et al., 2019; Gupta et al., 2019).”
BBN: Bilateral-Branch Network With Cumulative Learning for Long-Tailed Visual Recognition (2019, background): “, metric learning [34, 13], meta-learning [19] and knowledge transfer learning [28, 36], which, however, are not within the scope of this paper.”
Deep learning with noisy labels: exploring techniques and remedies in medical image analysis (2019, methodology): “Training procedures [102] - face recognition”
Global-Local GCN: Large-Scale Label Noise Cleansing for Face Recognition (2020, background): “Many studies [7, 20, 43, 44, 48, 58] point out that a noisy dataset is very harmful to model training.”
Forest R-CNN: Large-Vocabulary Long-Tailed Object Detection and Instance Segmentation (2020, background): “g. In addition to data re-balancing, many great works have been made using different ways,e.g., model finetuning [8, 29], metric learning [19, 44], meta-learning [27], and knowledge transfer learning [45]. Recently, [29, 31, 35, 36] also strive to solve the problem of object recognition in a large v”
Distribution Alignment: A Unified Framework for Long-tail Visual Recognition (2021, background): “Other work aim to enhance the representation or classifier head of tail categories by transferring knowledge from the head classes [41, 40, 26, 46, 8, 43, 42].”
Exploring Balanced Feature Spaces for Representation Learning (2021, background): “Such imbalanced datasets are very challenging for supervised learning methods to model, leading to noticeable performance drop (Wang et al., 2017; Mahajan et al., 2018; Zhong et al., 2019).”
Sub-center ArcFace: Boosting Face Recognition by Large-Scale Noisy Web Faces (2020, background): “However, computing time-varying weights for samples [11] or designing piece-wise loss functions [41] based on the current model’s predictions can only alleviate the influence from noisy data to some extent as the robustness and improvement depend on the initial performance of the model.”
Prime-Aware Adaptive Distillation (2020, background): “Many researchers[14,41] extended this idea to face recognition for noisy-robust feature learning.”