Modern neural networks have the capacity to overfit noisy labels frequently found in real-world datasets. Although great progress has been made, existing techniques are limited in providing theoretical guarantees for the performance of the neural networks trained with noisy labels. Here we propose a novel approach with strong theoretical guarantees for robust training of deep networks trained with noisy labels. The key idea behind our method is to select weighted subsets (coresets) of clean data points that provide an approximately low-rank Jacobian matrix. We then prove that gradient descent applied to the subsets do not overfit the noisy labels. Our extensive experiments corroborate our theory and demonstrate that deep networks trained on our subsets achieve a significantly superior performance compared to state-of-the art, e.g., 6% increase in accuracy on CIFAR-10 with 80% noisy labels, and 7% increase in accuracy on mini Webvision.
Group-invariant Coresets for Data-efficient Active Learning
Universidade Federal de Santa Catarina（圣卡塔琳娜联邦大学）；Universidade Católica de Pelotas（佩洛塔斯天主教大学）；Université de Lorraine, CNRS, CRAN（洛林大学，法国国家科学研究中心，自动控制与信号处理研究中心）
Middle East Technical University（中东技术大学）；ASELSAN（阿瑟尔桑公司）
The existence of noisy labels in the dataset causes significant performance degradation for deep neural networks (DNNs). To address this problem, we propose a Meta Soft Label Generation algorithm called MSLG, which can jointly generate soft labels using meta-learning techniques and learn DNN parameters in an end-to-end fashion. Our approach adapts the meta-learning paradigm to estimate optimal label distribution by checking gradient directions on both noisy training data and noise-free meta-data. In order to iteratively update soft labels, meta-gradient descent step is performed on estimated l…
University of California, Santa Cruz（加州大学圣克鲁兹分校）；Beijing University of Posts and Telecommunications（北京邮电大学）
The label noise transition matrix, characterizing the probabilities of a training instance being wrongly annotated, is crucial to designing popular solutions to learning with noisy labels. Existing works heavily rely on finding "anchor points" or their approximates, defined as instances belonging to a particular class almost surely. Nonetheless, finding anchor points remains a non-trivial task, and the estimation accuracy is also often throttled by the number of available anchor points. In this paper, we propose an alternative option to the above task. Our main contribution is the discovery of…
A $k$-decision tree $t$ (or $k$-tree) is a recursive partition of a matrix (2D-signal) into $k\geq 1$ block matrices (axis-parallel rectangles, leaves) where each rectangle is assigned a real label. Its regression or classification loss to a given matrix $D$ of $N$ entries (labels) is the sum of squared differences over every label in $D$ and its assigned label by $t$. Given an error parameter $\varepsilon\in(0,1)$, a $(k,\varepsilon)$-coreset $C$ of $D$ is a small summarization that provably approximates this loss to \emph{every} such tree, up to a multiplicative factor of $1\pm\varepsilon$. …
PICore: Physics-Informed Unsupervised Coreset Selection for Data Efficient Neural Operator Training
Department of Computer Science（计算机科学系）；University of Maryland, College Park（马里兰大学学院公园分校）；Georgia Institute of Technology（佐治亚理工学院）；Department of Mathematics（数学系）；Center for Scientific Computation and Mathematical Modeling（科学计算与数学建模中心）
Adobe Research（奥多比研究院）；University of Massachusetts Amherst（马萨诸塞大学阿默斯特分校）
We give relative error coresets for training linear classifiers with a broad class of loss functions, including the logistic loss and hinge loss. Our construction achieves $(1\pm ε)$ relative error with $\tilde O(d \cdot μ_y(X)^2/ε^2)$ points, where $μ_y(X)$ is a natural complexity measure of the data matrix $X \in \mathbb{R}^{n \times d}$ and label vector $y \in \{-1,1\}^n$, introduced in by Munteanu et al. 2018. Our result is based on subsampling data points with probabilities proportional to their $\ell_1$ $Lewis$ $weights$. It significantly improves on existing theoretical bounds and perfo…
University of Maryland（马里兰大学）；Universidad Simón Bolívar（西蒙·玻利瓦尔大学）
Given a training set $P$ of labeled points, the nearest-neighbor rule predicts the class of an unlabeled query point as the label of its closest point in the set. To improve the time and space complexity of classification, a natural question is how to reduce the training set without significantly affecting the accuracy of the nearest-neighbor rule. Nearest-neighbor condensation deals with finding a subset $R \subseteq P$ such that for every point $p \in P$, $p$'s nearest-neighbor in $R$ has the same label as $p$. This relates to the concept of coresets, which can be broadly defined as subsets …
Semi-supervised learning (SSL) algorithms have had great success in recent years in limited labeled data regimes. However, the current state-of-the-art SSL algorithms are computationally expensive and entail significant compute time and energy requirements. This can prove to be a huge limitation for many smaller companies and academic groups. Our main insight is that training on a subset of unlabeled data instead of entire unlabeled data enables the current SSL algorithms to converge faster, significantly reducing computational costs. In this work, we propose RETRIEVE, a coreset selection fram…
The Hong Kong University of Science and Technology（香港科技大学）；Google Research（谷歌研究院）
The goal of coreset selection in supervised learning is to produce a weighted subset of data, so that training only on the subset achieves similar performance as training on the entire dataset. Existing methods achieved promising results in resource-constrained scenarios such as continual learning and streaming. However, most of the existing algorithms are limited to traditional machine learning models. A few algorithms that can handle large models adopt greedy search approaches due to the difficulty in solving the discrete subset selection problem, which is computationally costly when coreset…
Rethinking Dataset Distillation: Hard Truths about Soft Labels
Vision and AI Lab, Indian Institute of Science, Bangalore（印度科学研究院视觉与人工智能实验室，班加罗尔）；University of Maryland, Baltimore County（马里兰大学巴尔的摩县分校）；IIT Hyderabad（海得拉巴理工学院）
GraphSculptor: Sculpting Pre-training Coreset for Graph Self-supervised Learning
Sangfor Technologies Inc.（Sangfor 技术公司）；Wuhan University（武汉大学）；The University of Melbourne（墨尔本大学）；Dalian Maritime University（大连海事大学）；Hunan University of Technology and Business（湖南工业大学；湘江实验室）；Xiangjiang Laboratory
Multi-label classification (MLC) is a generalization of standard classification where multiple labels may be assigned to a given sample. In the real world, it is more common to deal with noisy datasets than clean datasets, given how modern datasets are labeled by a large group of annotators on crowdsourcing platforms, but little attention has been given to evaluating multi-label classifiers with noisy labels. Exploiting label correlations now becomes a standard component of a multi-label classifier to achieve competitive performance. However, this component makes the classifier more prone to p…
Recently deep neural networks have shown their capacity to memorize training data, even with noisy labels, which hurts generalization performance. To mitigate this issue, we provide a simple but effective baseline method that is robust to noisy labels, even with severe noise. Our objective involves a variance regularization term that implicitly penalizes the Jacobian norm of the neural network on the whole training set (including the noisy-labeled data), which encourages generalization and prevents overfitting to the corrupted labels. Experiments on both synthetically generated incorrect label…
Leveraging weak or noisy supervision for building effective machine learning models has long been an important research problem. Its importance has further increased recently due to the growing need for large-scale datasets to train deep learning models. Weak or noisy supervision could originate from multiple sources including non-expert annotators or automatic labeling based on heuristics or user interaction signals. There is an extensive amount of previous work focusing on leveraging noisy labels. Most notably, recent work has shown impressive gains by using a meta-learned instance re-weight…
TU Delft（代尔夫特理工大学）；ABB Future Labs（ABB未来实验室）；Beijing Institute of Technology（北京理工大学）
Today's available datasets in the wild, e.g., from social media and open platforms, present tremendous opportunities and challenges for deep learning, as there is a significant portion of tagged images, but often with noisy, i.e. erroneous, labels. Recent studies improve the robustness of deep models against noisy labels without the knowledge of true labels. In this paper, we advocate to derive a stronger classifier which proactively makes use of the noisy labels in addition to the original images - turning noisy labels into learning features. To such an end, we propose a novel framework, Expe…
University of Sydney（悉尼大学）；Xidian University（西安电子科技大学）；RIKEN（理化学研究所）；Hong Kong Baptist University（香港浸会大学）；The University of Melbourne（墨尔本大学）；Brain-Inspired Technology Co., Ltd.（脑启发科技有限公司）
A similarity label indicates whether two instances belong to the same class while a class label shows the class of the instance. Without class labels, a multi-class classifier could be learned from similarity-labeled pairwise data by meta classification learning. However, since the similarity label is less informative than the class label, it is more likely to be noisy. Deep neural networks can easily remember noisy data, leading to overfitting in classification. In this paper, we propose a method for learning from only noisy-similarity-labeled data. Specifically, to model the noise, we employ…
UiT The Arctic University of Norway（挪威北极大学）；Universidad Rey Juan Carlos（胡安·卡洛斯国王大学）
Noisy labeled data represent a rich source of information that often are easily accessible and cheap to obtain, but label noise might also have many negative consequences if not accounted for. How to fully utilize noisy labels has been studied extensively within the framework of standard supervised machine learning over a period of several decades. However, very little research has been conducted on solving the challenge posed by noisy labels in non-standard settings. This includes situations where only a fraction of the samples are labeled (semi-supervised) and each high-dimensional sample is…
National Key Laboratory for Novel Software Technology（计算机软件新技术国家重点实验室）；Nanjing University（南京大学）
Deep neural networks need large amounts of labeled data to achieve good performance. In real-world applications, labels are usually collected from non-experts such as crowdsourcing to save cost and thus are noisy. In the past few years, deep learning methods for dealing with noisy labels have been developed, many of which are based on the small-loss criterion. However, there are few theoretical analyses to explain why these methods could learn well from noisy labels. In this paper, we theoretically explain why the widely-used small-loss criterion works. Based on the explanation, we reformalize…
The University of Sydney（悉尼大学）；Hong Kong Baptist University（香港浸会大学）；The University of Melbourne（墨尔本大学）；Xidian University（西安电子科技大学）；Brain-Inspired Technology Co., Ltd.（脑启发技术有限公司）；RIKEN Center for Advanced Intelligence Project（理化学研究所先进智能项目中心）
Learning with noisy labels has attracted a lot of attention in recent years, where the mainstream approaches are in pointwise manners. Meanwhile, pairwise manners have shown great potential in supervised metric learning and unsupervised contrastive learning. Thus, a natural question is raised: does learning in a pairwise manner mitigate label noise? To give an affirmative answer, in this paper, we propose a framework called Class2Simi: it transforms data points with noisy class labels to data pairs with noisy similarity labels, where a similarity label denotes whether a pair shares the class l…
University of California Santa Cruz（加州大学圣克鲁兹分校）；University of Sydney（悉尼大学）；RIKEN（理化学研究所）
Existing research on learning with noisy labels mainly focuses on synthetic label noise. Synthetic noise, though has clean structures which greatly enabled statistical analyses, often fails to model real-world noise patterns. The recent literature has observed several efforts to offer real-world noisy datasets, yet the existing efforts suffer from two caveats: (1) The lack of ground-truth verification makes it hard to theoretically study the property and treatment of real-world label noise; (2) These efforts are often of large scales, which may result in unfair comparisons of robust methods wi…
The Chinese University of Hong Kong（香港中文大学）；Tencent（腾讯）
Noisy labels are ubiquitous in real-world datasets, which poses a challenge for robustly training deep neural networks (DNNs) since DNNs can easily overfit to the noisy labels. Most recent efforts have been devoted to defending noisy labels by discarding noisy samples from the training set or assigning weights to training samples, where the weight associated with a noisy sample is expected to be small. Thereby, these previous efforts result in a waste of samples, especially those assigned with small weights. The input $x$ is always useful regardless of whether its observed label $y$ is clean. …
École Polytechnique（巴黎综合理工学院）；University of California, Berkeley（加州大学伯克利分校）
Class labels are often imperfectly observed, due to mistakes and to genuine ambiguity among classes. We propose a new semi-supervised deep generative model that explicitly models noisy labels, called the Mislabeled VAE (M-VAE). The M-VAE can perform better than existing deep generative models which do not account for label noise. Additionally, the derivation of M-VAE gives new theoretical insights into the popular M1+M2 semi-supervised model.
The Chinese University of Hong Kong（香港中文大学）；Tencent Technology（腾讯科技）
Noisy labels are ubiquitous in real-world datasets, which poses a challenge for robustly training deep neural networks (DNNs) as DNNs usually have the high capacity to memorize the noisy labels. In this paper, we find that the test accuracy can be quantitatively characterized in terms of the noise ratio in datasets. In particular, the test accuracy is a quadratic function of the noise ratio in the case of symmetric noise, which explains the experimental findings previously published. Based on our analysis, we apply cross-validation to randomly split noisy datasets, which identifies most sample…
University of Technology Sydney（悉尼科技大学）；RIKEN（理化学研究所）；Stanford University（斯坦福大学）；University of Tokyo（东京大学）；Paradigm Inc.（第四范式公司）
Deep learning with noisy labels is practically challenging, as the capacity of deep models is so high that they can totally memorize these noisy labels sooner or later during training. Nonetheless, recent studies on the memorization effects of deep neural networks show that they would first memorize training data of clean labels and then those of noisy labels. Therefore in this paper, we propose a new deep learning paradigm called Co-teaching for combating with noisy labels. Namely, we train two deep neural networks simultaneously, and let them teach each other given every mini-batch: firstly,…
We study the robustness to symmetric label noise of GNNs training procedures. By combining the nonlinear neural message-passing models (e.g. Graph Isomorphism Networks, GraphSAGE, etc.) with loss correction methods, we present a noise-tolerant approach for the graph classification task. Our experiments show that test accuracy can be improved under the artificial symmetric noisy setting.
Training of Convolutional Neural Networks (CNNs) with data with noisy labels is known to be a challenge. Based on the fact that directly providing the label to the data (Positive Learning; PL) has a risk of allowing CNNs to memorize the contaminated labels for the case of noisy data, the indirect learning approach that uses complementary labels (Negative Learning for Noisy Labels; NLNL) has proven to be highly effective in preventing overfitting to noisy data as it reduces the risk of providing faulty target. NLNL further employs a three-stage pipeline to improve convergence. As a result, filt…
Modern machine learning models are often trained on examples with noisy labels that hurt performance and are hard to identify. In this paper, we provide an empirical study showing that a simple $k$-nearest neighbor-based filtering approach on the logit layer of a preliminary model can remove mislabeled training data and produce more accurate models than many recently proposed methods. We also provide new statistical guarantees into its efficacy.
Deep neural networks (DNNs) trained on large-scale datasets have exhibited significant performance in image classification. Many large-scale datasets are collected from websites, however they tend to contain inaccurate labels that are termed as noisy labels. Training on such noisy labeled datasets causes performance degradation because DNNs easily overfit to noisy labels. To overcome this problem, we propose a joint optimization framework of learning DNN parameters and estimating true labels. Our framework can correct labels during training by alternating update of network parameters and label…
Bosch Center for Artificial Intelligence（博世人工智能中心）；BCAI（博世人工智能中心）；University of Ulm（乌尔姆大学）
Deep neural networks (DNNs) are powerful tools in computer vision tasks. However, in many realistic scenarios label noise is prevalent in the training images, and overfitting to these noisy labels can significantly harm the generalization performance of DNNs. We propose a novel technique to identify data with noisy labels based on the different distributions of the predictive uncertainties from a DNN over the clean and noisy data. Additionally, the behavior of the uncertainty over the course of training helps to identify the network weights which best can be used to relabel the noisy labels. D…
The ability of learning from noisy labels is very useful in many visual recognition tasks, as a vast amount of data with noisy labels are relatively easy to obtain. Traditionally, the label noises have been treated as statistical outliers, and approaches such as importance re-weighting and bootstrap have been proposed to alleviate the problem. According to our observation, the real-world noisy labels exhibit multi-mode characteristics as the true labels, rather than behaving like independent random outliers. In this work, we propose a unified distillation framework to use side information, inc…
We consider the learning from noisy labels (NL) problem which emerges in many real-world applications. In addition to the widely-studied synthetic noise in the NL literature, we also consider the pseudo labels in semi-supervised learning (Semi-SL) as a special case of NL. For both types of noise, we argue that the generalization performance of existing methods is highly coupled with the quality of noisy labels. Therefore, we counter the problem from a novel and unified perspective: learning from the auxiliary clean labels. Specifically, we propose the Rotational-Decoupling Consistency Regulari…
National University of Singapore（新加坡国立大学）；University of Minnesota（明尼苏达大学）
Despite the success of deep neural networks (DNNs) in image classification tasks, the human-level performance relies on massive training data with high-quality manual annotations, which are expensive and time-consuming to collect. There exist many inexpensive data sources on the web, but they tend to contain inaccurate labels. Training on noisy labeled datasets causes performance degradation because DNNs can easily overfit to the label noise. To overcome this problem, we propose a noise-tolerant training algorithm, where a meta-learning update is performed prior to conventional gradient update…
University of Maryland（马里兰大学）；Massachusetts Institute of Technology（麻省理工学院）；University of Toronto（多伦多大学）
Noisy labels are inevitable in large real-world datasets. In this work, we explore an area understudied by previous works -- how the network's architecture impacts its robustness to noisy labels. We provide a formal framework connecting the robustness of a network to the alignments between its architecture and target/noise functions. Our framework measures a network's robustness via the predictive power in its representations -- the test performance of a linear model trained on the learned representations using a small set of clean labels. We hypothesize that a network is more robust to noisy …
The University of Melbourne（墨尔本大学）；Tsinghua University（清华大学）；National Institute of Informatics（国立信息学研究所）
Datasets with significant proportions of noisy (incorrect) class labels present challenges for training accurate Deep Neural Networks (DNNs). We propose a new perspective for understanding DNN generalization for such datasets, by investigating the dimensionality of the deep representation subspace of training samples. We show that from a dimensionality perspective, DNNs exhibit quite distinctive learning styles when trained with clean labels versus when trained with a proportion of noisy labels. Based on this finding, we develop a new dimensionality-driven learning strategy, which monitors the…
Pohang University of Science and Technology（浦项科技大学）；Seoul National University of Science and Technology (SeoulTech)（首尔国立科技大学（首尔科技大学））
Because deep learning is vulnerable to noisy labels, sample selection techniques, which train networks with only clean labeled data, have attracted a great attention. However, if the labels are dominantly corrupted by few classes, these noisy samples are called dominant-noisy-labeled samples, the network also learns dominant-noisy-labeled samples rapidly via content-aware optimization. In this study, we propose a compelling criteria to penalize dominant-noisy-labeled samples intensively through class-wise penalty labels. By averaging prediction confidences for the each observed label, we obtai…
NAVER AI Lab（NAVER人工智能实验室）；Korea Advanced Institute of Science and Technology（韩国科学技术院）
Deep learning has achieved remarkable success in numerous domains with help from large amounts of big data. However, the quality of data labels is a concern because of the lack of high-quality labels in many real-world scenarios. As noisy labels severely degrade the generalization performance of deep neural networks, learning from noisy labels (robust training) is becoming an important task in modern deep learning applications. In this survey, we first describe the problem of learning with label noise from a supervised learning perspective. Next, we provide a comprehensive review of 62 state-o…
Most studies on learning from noisy labels rely on unrealistic models of i.i.d. label noise, such as class-conditional transition matrices. More recent work on instance-dependent noise models are more realistic, but assume a single generative process for label noise across the entire dataset. We propose a more principled model of label noise that generalizes instance-dependent noise to multiple labelers, based on the observation that modern datasets are typically annotated using distributed crowdsourcing methods. Under our labeler-dependent model, label noise manifests itself under two modalit…
Google Research（谷歌研究院）；Google Cloud AI（谷歌云人工智能）；Cornell University（康奈尔大学）
Performing controlled experiments on noisy data is essential in understanding deep learning across noise levels. Due to the lack of suitable datasets, previous research has only examined deep learning on controlled synthetic label noise, and real-world label noise has never been studied in a controlled setting. This paper makes three contributions. First, we establish the first benchmark of controlled real-world label noise from the web. This new benchmark enables us to study the web label noise in a controlled setting for the first time. The second contribution is a simple but effective metho…
University of Tokyo（东京大学）；NTT Communication Science Laboratories（NTT通信科学实验室）
Positive-unlabeled learning refers to the process of training a binary classifier using only positive and unlabeled data. Although unlabeled data can contain positive data, all unlabeled data are regarded as negative data in existing positive-unlabeled learning methods, which resulting in diminishing performance. We provide a new perspective on this problem -- considering unlabeled data as noisy-labeled data, and introducing a new formulation of PU learning as a problem of joint optimization of noisy-labeled data. This research presents a methodology that assigns initial pseudo-labels to unlab…
One-step abductive multi-target learning (OSAMTL) was proposed to handle complex noisy labels. In this paper, giving definition of diverse noisy label samples (DNLS), we propose one-step abductive multi-target learning with DNLS (OSAMTL-DNLS) to expand the methodology of original OSAMTL to better handle complex noisy labels.
Towards the Identifiability in Noisy Label Learning: A Multinomial Mixture Modelling Approach
Centre for Vision, Speech and Signal Processing, University of Surrey, United Kingdom（视觉、语音和信号处理中心，萨里大学，英国）；Department of Data Science and AI, Monash University, Australia（数据科学与人工智能系，莫纳什大学，澳大利亚）
The rawly collected training data often comes with separate noisy labels collected from multiple imperfect annotators (e.g., via crowdsourcing). A typical way of using these separate labels is to first aggregate them into one and apply standard training methods. The literature has also studied extensively on effective aggregation approaches. This paper revisits this choice and aims to provide an answer to the question of whether one should aggregate separate noisy labels into single ones or use them separately as given. We theoretically analyze the performance of both approaches under the empi…
The Pennsylvania State University（宾夕法尼亚州立大学）；Michigan State University（密歇根州立大学）
Graph Neural Networks (GNNs) have shown their great ability in modeling graph structured data. However, real-world graphs usually contain structure noises and have limited labeled nodes. The performance of GNNs would drop significantly when trained on such graphs, which hinders the adoption of GNNs on many applications. Thus, it is important to develop noise-resistant GNNs with limited labeled nodes. However, the work on this is rather limited. Therefore, we study a novel problem of developing robust GNNs on noisy graphs with limited labeled nodes. Our analysis shows that both the noisy edges …
Annotating the dataset with high-quality labels is crucial for performance of deep network, but in real world scenarios, the labels are often contaminated by noise. To address this, some methods were proposed to automatically split clean and noisy labels, and learn a semi-supervised learner in a Learning with Noisy Labels (LNL) framework. However, they leverage a handcrafted module for clean-noisy label splitting, which induces a confirmation bias in the semi-supervised learning phase and limits the performance. In this paper, we for the first time present a learnable module for clean-noisy la…
Massachusetts Institute of Technology（麻省理工学院）；South China University of Technology（华南理工大学）；Takeda Pharmaceuticals Co., Ltd.（武田药品工业株式会社）
Data lies at the core of modern deep learning. The impressive performance of supervised learning is built upon a base of massive accurately labeled data. However, in some real-world applications, accurate labeling might not be viable; instead, multiple noisy labels (instead of one accurate label) are provided by several annotators for each data sample. Learning a classifier on such a noisy training dataset is a challenging task. Previous approaches usually assume that all data samples share the same set of parameters related to annotator errors, while we demonstrate that label error learning s…
University of California, Santa Cruz（加州大学圣克鲁兹分校）；Brown University（布朗大学）；The University of Sydney（悉尼大学）；RIKEN AIP（理化学研究所人工智能研究中心）；University of Tokyo（东京大学）
Label smoothing (LS) is an arising learning paradigm that uses the positively weighted average of both the hard training labels and uniformly distributed soft labels. It was shown that LS serves as a regularizer for training data with hard labels and therefore improves the generalization of the model. Later it was reported LS even helps with improving robustness when learning with noisy labels. However, we observed that the advantage of LS vanishes when we operate in a high label noise regime. Intuitively speaking, this is due to the increased entropy of $\mathbb{P}(\text{noisy label}|X)$ when…
