Deep classifiers have achieved great success in visual recognition. However, real-world data is long-tailed by nature, leading to the mismatch between training and testing distributions. In this report, we introduce Balanced Activation (Balanced Softmax and Balanced Sigmoid), an elegant unbiased, and simple extension of Sigmoid and Softmax activation function, to accommodate the label distribution shift between training and testing in object detection. We derive the generalization bound for multiclass Softmax regression and show our loss minimizes the bound. In our experiments, we demonstrate that Balanced Activation generally provides ~3% gain in terms of mAP on LVIS-1.0 and outperforms the current state-of-the-art methods without introducing any extra parameters.
SenseTime Research（商汤科技研究院）；Nanyang Technological University（南洋理工大学）；National University of Singapore（新加坡国立大学）；The Chinese University of Hong Kong（香港中文大学）
Deep classifiers have achieved great success in visual recognition. However, real-world data is long-tailed by nature, leading to the mismatch between training and testing distributions. In this paper, we show that the Softmax function, though used in most classification tasks, gives a biased gradient estimation under the long-tailed setup. This paper presents Balanced Softmax, an elegant unbiased extension of Softmax, to accommodate the label distribution shift between training and testing. Theoretically, we derive the generalization bound for multiclass Softmax regression and show our loss m…
We consider infinitely-wide multi-layer perceptrons (MLPs) which are limits of standard deep feed-forward neural networks. We assume that, for each layer, the weights of an MLP are initialized with i.i.d. samples from either a light-tailed (finite variance) or heavy-tailed distribution in the domain of attraction of a symmetric $α$-stable distribution, where $α\in(0,2]$ may depend on the layer. For the bias terms of the layer, we assume i.i.d. initializations with a symmetric $α$-stable distribution having the same $α$ parameter of that layer. We then extend a recent result of Favaro, Fortini,…
Class imbalance is a ubiquitous phenomenon occurring in real world data distributions. To overcome its detrimental effect on training accurate classifiers, existing work follows three major directions: class re-balancing, information transfer, and representation learning. In this paper, we propose a new and complementary direction for improving performance on long tailed datasets - optimizing the backbone architecture through neural architecture search (NAS). We find that an architecture's accuracy obtained on a balanced dataset is not indicative of good performance on imbalanced ones. This po…
Adaptive Parametric Activation: Unifying and Generalising Activation Functions Across Tasks
Huawei Noah’s Ark Lab（华为诺亚实验室）；Department of Computing, Imperial College London（帝国理工学院计算机系）；Department of Computer Science, University of Liverpool（利物浦大学计算机科学系）；Department of Engineering, King’s College London（伦敦国王学院工程系）
Huazhong University of Science and Technology（华中科技大学）
The real-world data distribution is essentially long-tailed, which poses great challenge to the deep model. In this work, we propose a new method, Gradual Balanced Loss and Adaptive Feature Generator (GLAG) to alleviate imbalance. GLAG first learns a balanced and robust feature model with Gradual Balanced Loss, then fixes the feature model and augments the under-represented tail classes on the feature level with the knowledge from well-represented head classes. And the generated samples are mixed up with real training samples during training epochs. Gradual Balanced Loss is a general loss and …
University of Central Florida（中佛罗里达大学）；University of California at Merced（加州大学默塞德分校）；Google（谷歌）
Object frequency in the real world often follows a power law, leading to a mismatch between datasets with long-tailed class distributions seen by a machine learning model and our expectation of the model to perform well on all classes. We analyze this mismatch from a domain adaptation point of view. First of all, we connect existing class-balanced methods for long-tailed classification to target shift, a well-studied scenario in domain adaptation. The connection reveals that these methods implicitly assume that the training data and test data share the same class-conditioned distribution, whic…
ARC Centre of Excellence for Robotic Vision（ARC机器人视觉卓越中心）；University of Adelaide（阿德莱德大学）
Deep feedforward neural networks with piecewise linear activations are currently producing the state-of-the-art results in several public datasets. The combination of deep learning models and piecewise linear activation functions allows for the estimation of exponentially complex functions with the use of a large number of subnetworks specialized in the classification of similar input examples. During the training process, these subnetworks avoid overfitting with an implicit regularization scheme based on the fact that they must share their parameters with other subnetworks. Using this framewo…
Real-world imagery is often characterized by a significant imbalance of the number of images per class, leading to long-tailed distributions. An effective and simple approach to long-tailed visual recognition is to learn feature representations and a classifier separately, with instance and class-balanced sampling, respectively. In this work, we introduce a new framework, by making the key observation that a feature representation learned with instance sampling is far from optimal in a long-tailed setting. Our main contribution is a new training method, referred to as Class-Balanced Distillati…
Max Planck Institute for Informatics（马克斯·普朗克信息学研究所）；Saarland Informatics Campus（萨尔兰信息学园区）；University of Maryland, College Park（马里兰大学帕克分校）；CISPA Helmholtz Center for Information Security（亥姆霍兹信息安全研究中心）
Deep learning enables impressive performance in image recognition using large-scale artificially-balanced datasets. However, real-world datasets exhibit highly class-imbalanced distributions, yielding two main challenges: relative imbalance amongst the classes and data scarcity for mediumshot or fewshot classes. In this work, we address the problem of long-tailed recognition wherein the training set is highly imbalanced and the test set is kept balanced. Differently from existing paradigms relying on data-resampling, cost-sensitive learning, online hard example mining, loss objective reshaping…
Extracting Clean and Balanced Subset for Noisy Long-tailed Classification
Shenzhen Research Institute of Big Data（深圳大数据研究院）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））；CSIRO’s Data61（联邦科学与工业研究组织数据61部门）；The University of Sydney（悉尼大学）；School of Artificial Intelligence, Jilin University（吉林大学人工智能学院）
Fudan University（复旦大学）；Shanghai Collaborative Innovation Center on Intelligent Visual Computing（上海智能视觉计算协同创新中心）；La Trobe University（乐卓博大学）
Real-world data typically follow a long-tailed distribution, where a few majority categories occupy most of the data while most minority categories contain a limited number of samples. Classification models minimizing cross-entropy struggle to represent and classify the tail classes. Although the problem of learning unbiased classifiers has been well studied, methods for representing imbalanced data are under-explored. In this paper, we focus on representation learning for imbalanced data. Recently, supervised contrastive learning has shown promising performance on balanced data recently. Howe…
Tsinghua University（清华大学）；The Chinese University of Hong Kong（香港中文大学）
We present a new loss function called Distribution-Balanced Loss for the multi-label recognition problems that exhibit long-tailed class distributions. Compared to conventional single-label classification problem, multi-label recognition problems are often more challenging due to two significant issues, namely the co-occurrence of labels and the dominance of negative labels (when treated as multiple binary classification problems). The Distribution-Balanced Loss tackles these issues through two key modifications to the standard binary cross-entropy loss: 1) a new way to re-balance the weights …
A Strong Balanced-Softmax Classifier-Retraining Baseline for Long-Tailed Recognition
CICATA Querétaro Instituto Politecnico Nacional（国立理工学院克雷塔罗中心研究与高级研究中心）；Facultad de Informática Universidad Autónoma de Querétaro（克雷塔罗自治大学信息学院）；Escuela Nacional de Estudios Superiores Universidad Nacional Autónoma de México（国立自治大学高等研究学校）
Existing Graph Neural Networks (GNNs) usually assume a balanced situation where both the class distribution and the node degree distribution are balanced. However, in real-world situations, we often encounter cases where a few classes (i.e., head class) dominate other classes (i.e., tail class) as well as in the node degree perspective, and thus naively applying existing GNNs eventually fall short of generalizing to the tail cases. Although recent studies proposed methods to handle long-tail situations on graphs, they only focus on either the class long-tailedness or the degree long-tailedness…
Federated Learning (FL) has become a popular distributed learning paradigm that involves multiple clients training a global model collaboratively in a data privacy-preserving manner. However, the data samples usually follow a long-tailed distribution in the real world, and FL on the decentralized and long-tailed data yields a poorly-behaved global model severely biased to the head classes with the majority of the training samples. To alleviate this issue, decoupled training has recently been introduced to FL, considering it has achieved promising results in centralized long-tailed learning by …
Prior-free Balanced Replay: Uncertainty-guided Reservoir Sampling for Long-Tailed Continual Learning
Shenzhen Research Institute of Big Data (SRIBD)（深圳大数据研究院）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））；The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；The Hong Kong University of Science and Technology（香港科技大学）；The Hong Kong Polytechnic University（香港理工大学）
We address the overlooked unbiasedness in existing long-tailed classification methods: we find that their overall improvement is mostly attributed to the biased preference of tail over head, as the test distribution is assumed to be balanced; however, when the test is as imbalanced as the long-tailed training data -- let the test respect Zipf's law of nature -- the tail bias is no longer beneficial overall because it hurts the head majorities. In this paper, we propose Cross-Domain Empirical Risk Minimization (xERM) for training an unbiased model to achieve strong performances on both test dis…
Long-Tailed Semi-Supervised Learning (LTSSL) aims to learn from class-imbalanced data where only a few samples are annotated. Existing solutions typically require substantial cost to solve complex optimization problems, or class-balanced undersampling which can result in information loss. In this paper, we present the TRAS (TRAnsfer and Share) to effectively utilize long-tailed semi-supervised data. TRAS transforms the imbalanced pseudo-label distribution of a traditional SSL model via a delicate function to enhance the supervisory signals for minority classes. It then transfers the distributi…
Generalised Balanced Softmax: A Finite-Data Perspective on Logit Adjustment for Long-Tailed Recognition
School of Computing and Mathematical Sciences, University of Leicester（莱斯特大学计算与数学科学学院）；Department of Physics and Astronomy, University of Leicester（莱斯特大学物理与天文学系）；Scientific Computing, Rutherford Appleton Laboratory, Science and Technology Facilities Council（科学技术设施委员会卢瑟福·阿普尔顿实验室科学计算部）；School of Computer Science and Informatics, De Montfort University（德蒙福特大学计算机科学与信息学院）；School of Automation and Information Engineering, Xi’an University of Technology（西安理工大学自动化与信息工程学院）
Score-based model research in the last few years has produced state of the art generative models by employing Gaussian denoising score-matching (DSM). However, the Gaussian noise assumption has several high-dimensional limitations, motivating a more concrete route toward even higher dimension PDF estimation in future. We outline this limitation, before extending the theory to a broader family of noising distributions -- namely, the generalised normal distribution. To theoretically ground this, we relax a key assumption in (denoising) score matching theory, demonstrating that distributions whic…
The Hong Kong University of Science and Technology（香港科技大学）；Huawei Noah’s Ark Lab（华为诺亚方舟实验室）
Although deep neural networks achieve tremendous success on various classification tasks, the generalization ability drops sheer when training datasets exhibit long-tailed distributions. One of the reasons is that the learned representations (i.e. features) from the imbalanced datasets are less effective than those from balanced datasets. Specifically, the learned representation under class-balanced distribution will present the Neural Collapse (NC) phenomena. NC indicates the features from the same category are close to each other and from different categories are maximally distant, showing a…
Unsupervised Contrastive Learning Using Out-Of-Distribution Data for Long-Tailed Dataset
Department of Electronic Engineering, Seoul National University of Science and Technology（电子工程系，首尔科学技术大学）；Department of Electrical and Information Engineering, Seoul National University of Science and Technology（电气与信息工程系，首尔科学技术大学）；School of Electrical and Electronics Engineering, Chung-Ang University（电气电子工程学院， Chung-Ang 大学）
Indian Institute of Science（印度科学学院）；Indian Institute of Technology Tirupati（蒂鲁伯蒂印度理工学院）
Generative Adversarial Networks (GANs) have swiftly evolved to imitate increasingly complex image distributions. However, majority of the developments focus on performance of GANs on balanced datasets. We find that the existing GANs and their training regimes which work well on balanced datasets fail to be effective in case of imbalanced (i.e. long-tailed) datasets. In this work we introduce a novel theoretically motivated Class Balancing regularizer for training GANs. Our regularizer makes use of the knowledge from a pre-trained classifier to ensure balanced learning of all the classes in the…
A traditional approach to initialization in deep neural networks (DNNs) is to sample the network weights randomly for preserving the variance of pre-activations. On the other hand, several studies show that during the training process, the distribution of stochastic gradients can be heavy-tailed especially for small batch sizes. In this case, weights and therefore pre-activations can be modeled with a heavy-tailed distribution that has an infinite variance but has a finite (non-integer) fractional moment of order $s$ with $s<2$. Motivated by this fact, we develop initialization schemes for ful…
UC Santa Cruz（加州大学圣克鲁兹分校）；Riken（理化学研究所）；University of Sydney（悉尼大学）；Michigan State University（密歇根州立大学）；University of Tokyo（东京大学）；ByteDance Research（字节跳动研究院）
Both long-tailed and noisily labeled data frequently appear in real-world applications and impose significant challenges for learning. Most prior works treat either problem in an isolated way and do not explicitly consider the coupling effects of the two. Our empirical observation reveals that such solutions fail to consistently improve the learning when the dataset is long-tailed with label noise. Moreover, with the presence of label noise, existing methods do not observe universal improvements across different sub-populations; in other words, some sub-populations enjoyed the benefits of impr…
Training on datasets with long-tailed distributions has been challenging for major recognition tasks such as classification and detection. To deal with this challenge, image resampling is typically introduced as a simple but effective approach. However, we observe that long-tailed detection differs from classification since multiple classes may be present in one image. As a result, image resampling alone is not enough to yield a sufficiently balanced distribution at the object level. We address object-level resampling by introducing an object-centric memory replay strategy based on dynamic, ep…
U.S. Naval Research Laboratory（美国海军研究实验室）；Naval Center for Applied Research in AI（海军人工智能应用研究中心）
The cross-entropy softmax loss is the primary loss function used to train deep neural networks. On the other hand, the focal loss function has been demonstrated to provide improved performance when there is an imbalance in the number of training samples in each class, such as in long-tailed datasets. In this paper, we introduce a novel cyclical focal loss and demonstrate that it is a more universal loss function than cross-entropy softmax loss or focal loss. We describe the intuition behind the cyclical focal loss and our experiments provide evidence that cyclical focal loss provides superior …
Neural Processing Research Center, Seoul National University（首尔大学神经处理研究中心）
Continual learning from a sequential stream of data is a crucial challenge for machine learning research. Most studies have been conducted on this topic under the single-label classification setting along with an assumption of balanced label distribution. This work expands this research horizon towards multi-label classification. In doing so, we identify unanticipated adversity innately existent in many multi-label datasets, the long-tailed distribution. We jointly address the two independently solved problems, Catastropic Forgetting and the long-tailed label distribution by first empirically …
Xiamen University（厦门大学）；Hong Kong Baptist University（香港浸会大学）
Federated learning provides a privacy guarantee for generating good deep learning models on distributed clients with different kinds of data. Nevertheless, dealing with non-IID data is one of the most challenging problems for federated learning. Researchers have proposed a variety of methods to eliminate the negative influence of non-IIDness. However, they only focus on the non-IID data provided that the universal class distribution is balanced. In many real-world applications, the universal class distribution is long-tailed, which causes the model seriously biased. Therefore, this paper studi…
