# Rethinking Robust Representation Learning Under Fine-grained Noisy Faces

arXiv: https://arxiv.org/abs/2208.04352 · HTML: https://mlsanity.com/paper/2208.04352/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 38 references that are in the map.
- First posted on arXiv 8 Aug 2022.

## Builds on (most cited first, 20 of 20)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [FaceNet: A Unified Embedding for Face Recognition and Clustering](https://mlsanity.com/paper/1503.03832/) (2015)
- [CosFace: Large Margin Cosine Loss for Deep Face Recognition](https://mlsanity.com/paper/1801.09414/) (2018)
- [SphereFace: Deep Hypersphere Embedding for Face Recognition](https://mlsanity.com/paper/1704.08063/) (2017)
- [MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognition](https://mlsanity.com/paper/1607.08221/) (2016)
- [CurricularFace: Adaptive Curriculum Learning Loss for Deep Face Recognition](https://mlsanity.com/paper/2004.00288/) (2020)
- [Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval](https://mlsanity.com/paper/2004.01804/) (2020)
- [WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognition](https://mlsanity.com/paper/2103.04098/) (2021)
- [Partial FC: Training 10 Million Identities on a Single Machine](https://mlsanity.com/paper/2010.05222/) (2020)
- [AdaCos: Adaptively Scaling Cosine Logits for Effectively Learning Deep Face Representations](https://mlsanity.com/paper/1905.00292/) (2019)
- [Revisiting the Sibling Head in Object Detector](https://mlsanity.com/paper/2003.07540/) (2020)
- [Mis-classified Vector Guided Softmax Loss for Face Recognition](https://mlsanity.com/paper/1912.00833/) (2019)
- [Region-based Quality Estimation Network for Large-scale Person Re-identification](https://mlsanity.com/paper/1711.08766/) (2017)
- [Noise-Tolerant Paradigm for Training Face Recognition CNNs](https://mlsanity.com/paper/1903.10357/) (2019)
- [Semi-Siamese Training for Shallow Face Learning](https://mlsanity.com/paper/2007.08398/) (2020)
- [Towards Flops-constrained Face Recognition](https://mlsanity.com/paper/1909.00632/) (2019)
- [Discriminability Distillation in Group Representation Learning](https://mlsanity.com/paper/2008.10850/) (2020)
- [3rd Place Solution to "Google Landmark Retrieval 2020"](https://mlsanity.com/paper/2008.10480/) (2020)
- [Beyond Trade-off: Accelerate FCN-based Face Detector with Higher Accuracy](https://mlsanity.com/paper/1804.05197/) (2018)

## Built on by (most cited first, 1 of 1)

- [High-Fidelity Diffusion Face Swapping with ID-Constrained Facial Conditioning](https://mlsanity.com/paper/2503.22179/) (2025)

## Abstract (arXiv)

Learning robust feature representation from large-scale noisy faces stands out as one of the key challenges in high-performance face recognition. Recent attempts have been made to cope with this challenge by alleviating the intra-class conflict and inter-class conflict. However, the unconstrained noise type in each conflict still makes it difficult for these algorithms to perform well. To better understand this, we reformulate the noise type of each class in a more fine-grained manner as N-identities|KC-clusters. Different types of noisy faces can be generated by adjusting the values of nkc. Based on this unified formulation, we found that the main barrier behind the noise-robust representation learning is the flexibility of the algorithm under different N, K, and C. For this potential problem, we propose a new method, named Evolving Sub-centers Learning~(ESL), to find optimal hyperplanes to accurately describe the latent space of massive noisy faces. More specifically, we initialize M sub-centers for each class and ESL encourages it to be automatically aligned to N-identities|KC-clusters faces via producing, merging, and dropping operations. Images belonging to the same identity in noisy faces can effectively converge to the same sub-center and samples with different identities will be pushed away. We inspect its effectiveness with an elaborate ablation study on the synthetic noisy dataset with different N, K, and C. Without any bells and whistles, ESL can achieve significant performance gains over state-of-the-art methods on large-scale noisy faces