Provably End-to-end Label-Noise Learning without Anchor Points
University of New South Wales（新南威尔士大学）；University of Sydney（悉尼大学）；Hong Kong Baptist University（香港浸会大学）；RIKEN AIP（日本理化学研究所创新智能综合研究中心）；University of Tokyo（东京大学）
Xuefeng Li, Tongliang Liu, Bo Han, Gang Niu, Masashi Sugiyama
In label-noise learning, the transition matrix plays a key role in building statistically consistent classifiers. Existing consistent estimators for the transition matrix have been developed by exploiting anchor points. However, the anchor-point assumption is not always satisfied in real scenarios. In this paper, we propose an end-to-end framework for solving label-noise learning without anchor points, in which we simultaneously optimize two objectives: the cross entropy loss between the noisy label and the predicted probability by the neural network, and the volume of the simplex formed by the columns of the transition matrix. Our proposed framework can identify the transition matrix if the clean class-posterior probabilities are sufficiently scattered. This is by far the mildest assumption under which the transition matrix is provably identifiable and the learned classifier is statistically consistent. Experimental results on benchmark datasets demonstrate the effectiveness and robustness of the proposed method.
University of Sydney（悉尼大学）；Xidian University（西安电子科技大学）；RIKEN（理化学研究所）；Nanjing University of Science and Technology（南京理工大学）；University of Tokyo（东京大学）
In label-noise learning, \textit{noise transition matrix}, denoting the probabilities that clean labels flip into noisy labels, plays a central role in building \textit{statistically consistent classifiers}. Existing theories have shown that the transition matrix can be learned by exploiting \textit{anchor points} (i.e., data points that belong to a specific class almost surely). However, when there are no anchor points, the transition matrix will be poorly learned, and those current consistent classifiers will significantly degenerate. In this paper, without employing anchor points, we propos…
University of Bristol（布里斯托大学）；Cookpad Ltd（Cookpad有限公司）
In this paper we provide machine learning practitioners with tools to answer the question: is there class-conditional noise in my labels? In particular, we present hypothesis tests to check whether a given dataset of instance-label pairs has been corrupted with class-conditional label noise, as opposed to uniform label noise, with the former biasing learning, while the latter -- under mild conditions -- does not. The outcome of these tests can then be used in conjunction with other information to assess further steps. While previous works explore the direct estimation of the noise rates, this …
School of Computer Science and Engineering, Nanyang Technological University, Singapore（南洋理工大学计算机科学与工程学院）；School of Computer Science and Engineering, Southeast University（东南大学计算机科学与工程学院）；Department of Computer Science, Hong Kong Baptist University（香港浸会大学计算机科学系）；The University of Queensland, Australia（昆士兰大学）；Center for Advanced Intelligence Project, RIKEN（理化学研究所高级智能项目中心）；The University of Tokyo, Japan（东京大学）
Partial-label learning (PLL) is a multi-class classification problem, where each training example is associated with a set of candidate labels. Even though many practical PLL methods have been proposed in the last two decades, there lacks a theoretical understanding of the consistency of those methods-none of the PLL methods hitherto possesses a generation process of candidate label sets, and then it is still unclear why such a method works on a specific dataset and when it may fail given a different dataset. In this paper, we propose the first generation model of candidate label sets, and dev…
King Abdullah University of Science and Technology（阿卜杜拉国王科技大学）；Norton Research Group（诺顿研究集团）
Evasion attack in multi-label learning systems is an interesting, widely witnessed, yet rarely explored research topic. Characterizing the crucial factors determining the attackability of the multi-label adversarial threat is the key to interpret the origin of the adversarial vulnerability and to understand how to mitigate it. Our study is inspired by the theory of adversarial risk bound. We associate the attackability of a targeted multi-label classifier with the regularity of the classifier and the training data distribution. Beyond the theoretical attackability analysis, we further propose …
University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Huazhong University of Science and Technology（华中科技大学）；Duke University（杜克大学）
Multi-label classification, which predicts a set of labels for an input, has many applications. However, multiple recent studies showed that multi-label classification is vulnerable to adversarial examples. In particular, an attacker can manipulate the labels predicted by a multi-label classifier for an input via adding carefully crafted, human-imperceptible perturbation to it. Existing provable defenses for multi-class classification achieve sub-optimal provable robustness guarantees when generalized to multi-label classification. In this work, we propose MultiGuard, the first provably robust…
University of California, Riverside（加州大学河滨分校）；University of Southern California（南加州大学）
Modern neural networks are typically trained in an over-parameterized regime where the parameters of the model far exceed the size of the training data. Such neural networks in principle have the capacity to (over)fit any set of labels including pure noise. Despite this, somewhat paradoxically, neural network models trained via first-order methods continue to predict well on yet unseen test data. This paper takes a step towards demystifying this phenomena. Under a rich dataset model, we show that gradient descent is provably robust to noise/corruption on a constant fraction of the labels despi…
Deep neural networks (DNNs) are capable of perfectly fitting the training data, including memorizing noisy data. It is commonly believed that memorization hurts generalization. Therefore, many recent works propose mitigation strategies to avoid noisy data or correct memorization. In this work, we step back and ask the question: Can deep learning be robust against massive label noise without any mitigation? We provide an affirmative answer for the case of symmetric label noise: We find that certain DNNs, including under-parameterized and over-parameterized models, can tolerate massive symmetric…
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…
Federated learning enables clients to collaboratively learn a shared global model without sharing their local training data with a cloud server. However, malicious clients can corrupt the global model to predict incorrect labels for testing examples. Existing defenses against malicious clients leverage Byzantine-robust federated learning methods. However, these methods cannot provably guarantee that the predicted label for a testing example is not affected by malicious clients. We bridge this gap via ensemble federated learning. In particular, given any base federated learning algorithm, we us…
South China University of Technology（华南理工大学）；Pazhou Lab（琶洲实验室）；DAMO Academy, Alibaba Group（阿里巴巴集团达摩院）；The Hong Kong Polytechnic University（香港理工大学）
Motivated by the problem relatedness between unsupervised domain adaptation (UDA) and semi-supervised learning (SSL), many state-of-the-art UDA methods adopt SSL principles (e.g., the cluster assumption) as their learning ingredients. However, they tend to overlook the very domain-shift nature of UDA. In this work, we take a step further to study the proper extensions of SSL techniques for UDA. Taking the algorithm of label propagation (LP) as an example, we analyze the challenges of adopting LP to UDA and theoretically analyze the conditions of affinity graph/matrix construction in order to a…
In overparametrized models, the noise in stochastic gradient descent (SGD) implicitly regularizes the optimization trajectory and determines which local minimum SGD converges to. Motivated by empirical studies that demonstrate that training with noisy labels improves generalization, we study the implicit regularization effect of SGD with label noise. We show that SGD with label noise converges to a stationary point of a regularized loss $L(θ) +λR(θ)$, where $L(θ)$ is the training loss, $λ$ is an effective regularization parameter depending on the step size, strength of the label noise, and the…
DSReg: Provably Recovering Individual World Latents without Reconstruction
University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Cold Spring Harbor Laboratory（冷泉港实验室）；Brown University（布朗大学）；Max Planck Institute for Intelligent Systems（马克斯·普朗克智能系统研究所）；ELLIS Institute Tübingen（ELLIS图宾根研究所）
MASA-TCN: Multi-anchor Space-aware Temporal Convolutional Neural Networks for Continuous and Discrete EEG Emotion Recognition
School of Computer Science and Engineering, Nanyang Technological University（南洋理工大学计算机科学与工程学院）；School of Biological Science and Medical Engineering, Southeast University（东南大学生物科学与医学工程学院）
We prove a new upper bound on the generalization gap of classifiers that are obtained by first using self-supervision to learn a representation $r$ of the training data, and then fitting a simple (e.g., linear) classifier $g$ to the labels. Specifically, we show that (under the assumptions described below) the generalization gap of such classifiers tends to zero if $\mathsf{C}(g) \ll n$, where $\mathsf{C}(g)$ is an appropriately-defined measure of the simple classifier $g$'s complexity, and $n$ is the number of training samples. We stress that our bound is independent of the complexity of the …
Variance-aware Reward Modeling with Anchor Guidance
Department of Data Science and Artificial Intelligence, Hong Kong Polytechnic University（香港理工大学数据科学与人工智能系）；School of Statistics and Data Science, Shanghai University of Finance and Economics（上海财经大学统计与数据科学学院）
3D point cloud classification has many safety-critical applications such as autonomous driving and robotic grasping. However, several studies showed that it is vulnerable to adversarial attacks. In particular, an attacker can make a classifier predict an incorrect label for a 3D point cloud via carefully modifying, adding, and/or deleting a small number of its points. Randomized smoothing is state-of-the-art technique to build certifiably robust 2D image classifiers. However, when applied to 3D point cloud classification, randomized smoothing can only certify robustness against adversarially m…
Stress-Testing Neural Network Verifiers with Provably Robust Instances
Department of Statistics & Data Science, University of California, Los Angeles（统计与数据科学系，加州大学洛杉矶分校）；Department of Mathematics, University of California, Los Angeles（数学系，加州大学洛杉矶分校）；Max Planck Institute for Mathematics in the Sciences, Leipzig（科学数学研究所，莱比锡）
The label noise transition matrix, denoting the transition probabilities from clean labels to noisy labels, is crucial for designing statistically robust solutions. Existing estimators for noise transition matrices, e.g., using either anchor points or clusterability, focus on computer vision tasks that are relatively easier to obtain high-quality representations. We observe that tasks with lower-quality features fail to meet the anchor-point or clusterability condition, due to the coexistence of both uninformative and informative representations. To handle this issue, we propose a generic and …
The Ohio State University（俄亥俄州立大学）；New York University（纽约大学）
Given labeled points in a high-dimensional vector space, we seek a low-dimensional subspace such that projecting onto this subspace maintains some prescribed distance between points of differing labels. Intended applications include compressive classification. Taking inspiration from large margin nearest neighbor classification, this paper introduces a semidefinite relaxation of this problem. Unlike its predecessors, this relaxation is amenable to theoretical analysis, allowing us to provably recover a planted projection operator from the data.
Xian Jiaotong University（西安交通大学）；Macau University of Science and Technology（澳门科技大学）
To discover intrinsic inter-class transition probabilities underlying data, learning with noise transition has become an important approach for robust deep learning on corrupted labels. Prior methods attempt to achieve such transition knowledge by pre-assuming strongly confident anchor points with 1-probability belonging to a specific class, generally infeasible in practice, or directly jointly estimating the transition matrix and learning the classifier from the noisy samples, always leading to inaccurate estimation misguided by wrong annotation information especially in large noise cases. To…
