DAIL: Dataset-Aware and Invariant Learning for Face Recognition

Wang, Gaoang; Chen, Lin; Liu, Tianqiang; He, Mingwei; Luo, Jiebo

doi:10.1109/icpr48806.2021.9412121

Cited by 5 publications

(2 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since the class distribution of the training set is usually imbalanced, existing algorithms in long-tailed learning attempt to improve the generalization ability of DNN towards the medium and tail classes. Generally, the framework of existing methods can be summarized into four categories: Resampling (Chawla et al 2002;Buda, Maki, and Mazurowski 2018) conducts under-sampling on head classes or oversampling on tail classes to balance the training data; Reweighting (Khan et al 2017;Cui et al 2019) adjusts the class weight or sample weight to maintain a balanced training process; Ensemble learning (Wang et al 2020;Cai, Wang, and Hwang 2021;Zhang et al 2022) is based on multiple experts (classifier heads) to enhance the representation learning; Loss modification (Cao et al 2019;Hong et al 2021) modifies the logit value by margins in either the training or inference stage and the typical work "logit adjustment" (Menon et al 2020) which has been proved as the Bayes-optimal solution of the long-tailed problem. Nonetheless, the current approaches are predominantly tailored for the fully-supervised context, rendering them unsuitable for semi-supervised learning where the class distribution is unspecified.…”

Section: Long-tailed Learningmentioning

confidence: 99%

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Ma,

Elezi,

Deng

et al. 2024

AAAI

View full text Add to dashboard Cite

We address the challenging problem of Long-Tailed Semi-Supervised Learning (LTSSL) where labeled data exhibit imbalanced class distribution and unlabeled data follow an unknown distribution. Unlike in balanced SSL, the generated pseudo-labels are skewed towards head classes, intensifying the training bias. Such a phenomenon is even amplified as more unlabeled data will be mislabeled as head classes when the class distribution of labeled and unlabeled datasets are mismatched. To solve this problem, we propose a novel method named ComPlementary Experts (CPE). Specifically, we train multiple experts to model various class distributions, each of them yielding high-quality pseudo-labels within one form of class distribution. Besides, we introduce Classwise Batch Normalization for CPE to avoid performance degradation caused by feature distribution mismatch between head and non-head classes. CPE achieves state-of-the-art performances on CIFAR-10-LT, CIFAR-100-LT, and STL-10-LT dataset benchmarks. For instance, on CIFAR-10-LT, CPE improves test accuracy by over >2.22% compared to baselines. Code is available at https://github.com/machengcheng2016/CPE-LTSSL.

show abstract

Section: Long-tailed Learningmentioning

confidence: 99%

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Ma,

Elezi,

Deng

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…The prevalent domain adaptation techniques for face recognition (FR) mainly fall into unsupervised domain adaptation (UDA) [16][17][18][19] . UDA suggests that global alignment between domains would induce per-class alignment, but massive training data must be compelling.…”

Section: Introductionmentioning

confidence: 99%

DIRA: disjoint-identity resolution adaptation for low-resolution face recognition

Chai

Low²,

Teoh³

2022

Fourteenth International Conference on Digital Image Processing (ICDIP 2022)

View full text Add to dashboard Cite

Low-resolution face recognition (LRFR) intends to identify unknown poor-quality face images and is widely employed in real-world surveillance applications. While collecting a large-scale labeled low-resolution (LR) face dataset could be conducive, it is practically infeasible due to labor costs and privacy issues. In contrast, accessing high-resolution (HR) face datasets is relatively effortless. However, prevailing domain adaptation techniques are often tenuous as they demand sharing of similar face images at different resolutions. We propose disjoint-identity resolution adaptation (DIRA) to transfer substantial face semantic representations from HR to LR face images, despite disjoint identities and limited labeled LR images. We accredit that continuous adversarial learning between HR-LR resolution alignment and segregation renders effective feature extraction and discriminative LR face representation. Our experimental results show a notable performance boost over the recent state-of-the-art methods for the challenging realistic low-resolution face recognition task.

show abstract

Age-invariant face network (AFN): a discriminative model towards age-invariant face recognition

Li,

Zhou,

Chen

2024

Neural Comput & Applic

View full text Add to dashboard Cite

DAIL: Dataset-Aware and Invariant Learning for Face Recognition

Cited by 5 publications

References 40 publications

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

Three Heads Are Better than One: Complementary Experts for Long-Tailed Semi-supervised Learning

DIRA: disjoint-identity resolution adaptation for low-resolution face recognition

Age-invariant face network (AFN): a discriminative model towards age-invariant face recognition

Contact Info

Product

Resources

About