Class Rectification Hard Mining for Imbalanced Deep Learning

By Qi Dong, Shaogang Gong, Xiatian Zhu. 2017

223 citations recorded by the source corpus.

Recognising detailed facial or clothing attributes in images of people is a challenging task for computer vision, especially when the training data are both in very large scale and extremely imbalanced among different attribute classes. To address this problem, we formulate a novel scheme for batch incremental hard sample mining of minority attribute classes from imbalanced large scale training data. We develop an end-to-end deep learning framework capable of avoiding the dominant effect of majority classes by discovering sparsely sampled boundaries of minority classes. This is made possible by introducing a Class Rectification Loss (CRL) regularising algorithm. We demonstrate the advantages and scalability of CRL over existing state-of-the-art attribute recognition and imbalanced data learning models on two large scale imbalanced benchmark datasets, the CelebA facial attribute dataset and the X-Domain clothing attribute dataset.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Large-Scale Long-Tailed Recognition in an Open World (2019, methodology): “Some recent methods include metric learning [22, 33], hard negative mining [10, 27], and meta learning [15, 55].”
Class-Balanced Loss Based on Effective Number of Samples (2019, background): “Another line of important work aims to study sample difficulty in terms of loss and assign higher weights to hard examples [11, 29, 8, 26].”
Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier (2020, background): “A typical approach is to weight classes by their frequency [34, 47] or treat tail classes as hard examples [19].”
Learning to Segment the Tail (2020, background): “The former aims to re-balance the training samples across classes [16, 3, 11, 6]; while the latter focuses on assigning different weights to adjust the loss function [18, 40, 47, 7].”
ELF: An Early-Exiting Framework for Long-Tailed Classification (2020, background): “Reweighting based approaches assign a larger weight to rare class examples and a smaller weight to the majority ones [6, 18, 19, 20, 21, 22].”
Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting (2019, methodology): “[41] Qi Dong, Shaogang Gong, and Xiatian Zhu.”
Learning a Unified Classifier Incrementally via Rebalancing (2019, background): “Among the previous works, the one presented in [9] is the most relevant to ours.”
Transferability and Hardness of Supervised Classification Tasks (2019, background): “[16] Qi Dong, Shaogang Gong, and Xiatian Zhu.”
Learning to Reweight with Deep Interactions (2021, background): “…motivation of data weighting is to increase the robustness of training, including addressing the problem of imbalanced data (Sun et al. 2007; Dong, Gong, and Zhu 2017; Khan et al. 2018), biased data (Zadrozny 2004; Ren et al. 2018), noisy data (Angluin and Laird 1988; Reed et al. 2014;…”