A model must adapt itself to generalize to new and different data during testing. In this setting of fully test-time adaptation the model has only the test data and its own parameters. We propose to adapt by test entropy minimization (tent): we optimize the model for confidence as measured by the entropy of its predictions. Our method estimates normalization statistics and optimizes channel-wise affine transformations to update online on each batch. Tent reduces generalization error for image classification on corrupted ImageNet and CIFAR-10/100 and reaches a new state-of-the-art error on ImageNet-C. Tent handles source-free domain adaptation on digit recognition from SVHN to MNIST/MNIST-M/USPS, on semantic segmentation from GTA to Cityscapes, and on the VisDA-C benchmark. These results are achieved in one epoch of test-time optimization without altering training.
Nanjing University of Posts and Telecommunications（南京邮电大学）；Southeast University（东南大学）；Temple University（天普大学）
Entropy minimization has been widely used in unsupervised domain adaptation (UDA). However, existing works reveal that entropy minimization only may result into collapsed trivial solutions. In this paper, we propose to avoid trivial solutions by further introducing diversity maximization. In order to achieve the possible minimum target risk for UDA, we show that diversity maximization should be elaborately balanced with entropy minimization, the degree of which can be finely controlled with the use of deep embedded validation in an unsupervised manner. The proposed minimal-entropy diversity ma…
We present a new algorithm for domain adaptation improving upon a discrepancy minimization algorithm previously shown to outperform a number of algorithms for this task. Unlike many previous algorithms for domain adaptation, our algorithm does not consist of a fixed reweighting of the losses over the training sample. We show that our algorithm benefits from a solid theoretical foundation and more favorable learning bounds than discrepancy minimization. We present a detailed description of our algorithm and give several efficient solutions for solving its optimization problem. We also report th…
In this paper, our goal is to adapt a pre-trained convolutional neural network to domain shifts at test time. We do so continually with the incoming stream of test batches, without labels. The existing literature mostly operates on artificial shifts obtained via adversarial perturbations of a test image. Motivated by this, we evaluate the state of the art on two realistic and challenging sources of domain shifts, namely contextual and semantic shifts. Contextual shifts correspond to the environment types, for example, a model pre-trained on indoor context has to adapt to the outdoor context on…
Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Peng Cheng Laboratory（鹏城实验室）；Huawei Technologies（华为技术有限公司）
Due to the domain discrepancy in visual domain adaptation, the performance of source model degrades when bumping into the high data density near decision boundary in target domain. A common solution is to minimize the Shannon Entropy to push the decision boundary away from the high density area. However, entropy minimization also leads to severe reduction of prediction diversity, and unfortunately brings harm to the domain adaptation. In this paper, we investigate the prediction discriminability and diversity by studying the structure of the classification output matrix of a randomly selected …
In this paper, we propose a new approach called Deep LogCORAL for unsupervised visual domain adaptation. Our work builds on the recently proposed Deep CORAL method, which proposed to train a convolutional neural network and simultaneously minimize the Euclidean distance of convariance matrices between the source and target domains. We propose to use the Riemannian distance, approximated by Log-Euclidean distance, to replace the naive Euclidean distance in Deep CORAL. We also consider first-order information, and minimize the distance of mean vectors between two domains. We build an end-to-end …
Semantic segmentation is a key problem for many computer vision tasks. While approaches based on convolutional neural networks constantly break new records on different benchmarks, generalizing well to diverse testing environments remains a major challenge. In numerous real world applications, there is indeed a large gap between data distributions in train and test domains, which results in severe performance loss at run-time. In this work, we address the task of unsupervised domain adaptation in semantic segmentation with losses based on the entropy of the pixel-wise predictions. To this end,…
University of Electronic Science and Technology of China（电子科技大学）
Unsupervised Domain Adaptation (UDA) aims to generalize the knowledge learned from a well-labeled source domain to an unlabeled target domain. Recently, adversarial domain adaptation with two distinct classifiers (bi-classifier) has been introduced into UDA which is effective to align distributions between different domains. Previous bi-classifier adversarial learning methods only focus on the similarity between the outputs of two distinct classifiers. However, the similarity of the outputs cannot guarantee the accuracy of target samples, i.e., target samples may match to wrong categories even…
National University of Singapore（新加坡国立大学）；Nanyang Technological University（南洋理工大学）
Reducing domain divergence is a key step in transfer learning problems. Existing works focus on the minimization of global domain divergence. However, two domains may consist of several shared subdomains, and differ from each other in each subdomain. In this paper, we take the local divergence of subdomains into account in transfer. Specifically, we propose to use low-dimensional manifold to represent subdomain, and align the local data distribution discrepancy in each manifold across domains. A Manifold Maximum Mean Discrepancy (M3D) is developed to measure the local distribution discrepancy …
Unsupervised domain adaptation aims to transfer knowledge from a labeled source domain to an unlabeled target domain. Previous methods focus on learning domain-invariant features to decrease the discrepancy between the feature distributions as well as minimizing the source error and have made remarkable progress. However, a recently proposed theory reveals that such a strategy is not sufficient for a successful domain adaptation. It shows that besides a small source error, both the discrepancy between the feature distributions and the discrepancy between the labeling functions should be small …
CyberAgent, Inc.（赛博代理公司）；The University of Tokyo（东京大学）
Density ratio estimation (DRE) is at the core of various machine learning tasks such as anomaly detection and domain adaptation. In existing studies on DRE, methods based on Bregman divergence (BD) minimization have been extensively studied. However, BD minimization when applied with highly flexible models, such as deep neural networks, tends to suffer from what we call train-loss hacking, which is a source of overfitting caused by a typical characteristic of empirical BD estimators. In this paper, to mitigate train-loss hacking, we propose a non-negative correction for empirical BD estimators…
Entropy Minimization without Model Collapse: Mitigating Prediction Bias in Medical Imaging
School of Computation, Information and Technology, Technical University of Munich, Germany（慕尼黑技术大学计算、信息与技术学院）；Institute of Machine Learning in Biomedical Imaging, Helmholtz Munich, Germany（生物医学成像中的机器学习研究所，海德堡慕尼黑德国）；School of Biomedical Engineering and Imaging Sciences, King’s College London, UK（伦敦国王学院生物医学工程与成像科学学院）；Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心（MCML））；relAI – Konrad Zuse School of Excellence in Reliable AI（relAI——Konrad Zuse可靠性人工智能卓越学院）；TUM University Hospital Rechts der Isar（慕尼黑技术大学医院Rechts der Isar）
University of Freiburg（弗赖堡大学）；Bosch Center for Artificial Intelligence（博世人工智能中心）
Deep neural networks often exhibit poor performance on data that is unlikely under the train-time data distribution, for instance data affected by corruptions. Previous works demonstrate that test-time adaptation to data shift, for instance using entropy minimization, effectively improves performance on such shifted distributions. This paper focuses on the fully test-time adaptation setting, where only unlabeled data from the target distribution is required. This allows adapting arbitrary pretrained networks. Specifically, we propose a novel loss that improves test-time adaptation by addressin…
Forensic audio analysis for speaker verification offers unique challenges due to location/scenario uncertainty and diversity mismatch between reference and naturalistic field recordings. The lack of real naturalistic forensic audio corpora with ground-truth speaker identity represents a major challenge in this field. It is also difficult to directly employ small-scale domain-specific data to train complex neural network architectures due to domain mismatch and loss in performance. Alternatively, cross-domain speaker verification for multiple acoustic environments is a challenging task which co…
Source-free domain adaptation (SFDA) aims to transfer knowledge learned from a source domain to an unlabeled target domain, where the source data is unavailable during adaptation. Existing approaches for SFDA focus on self-training usually including well-established entropy minimization techniques. One of the main challenges in SFDA is to reduce accumulation of errors caused by domain misalignment. A recent strategy successfully managed to reduce error accumulation by pseudo-labeling the target samples based on class-wise prototypes (centroids) generated by their clustering in the representati…
Domain adaptation addresses the common problem when the target distribution generating our test data drifts from the source (training) distribution. While absent assumptions, domain adaptation is impossible, strict conditions, e.g. covariate or label shift, enable principled algorithms. Recently-proposed domain-adversarial approaches consist of aligning source and target encodings, often motivating this approach as minimizing two (of three) terms in a theoretical bound on target error. Unfortunately, this minimization can cause arbitrary increases in the third term, e.g. they can break down un…
Zhejiang University（浙江大学）；Fabu Inc.（法布公司）；Alibaba-Zhejiang University Joint Institute of Frontier Technologies（阿里巴巴-浙江大学前沿技术联合研究院）
Deep neural networks for semantic segmentation always require a large number of samples with pixel-level labels, which becomes the major difficulty in their real-world applications. To reduce the labeling cost, unsupervised domain adaptation (UDA) approaches are proposed to transfer knowledge from labeled synthesized datasets to unlabeled real-world datasets. Recently, some semi-supervised learning methods have been applied to UDA and achieved state-of-the-art performance. One of the most popular approaches in semi-supervised learning is the entropy minimization method. However, when applying …
In many real-world applications, we want to exploit multiple source datasets of similar tasks to learn a model for a different but related target dataset -- e.g., recognizing characters of a new font using a set of different fonts. While most recent research has considered ad-hoc combination rules to address this problem, we extend previous work on domain discrepancy minimization to develop a finite-sample generalization bound, and accordingly propose a theoretically justified optimization procedure. The algorithm we develop, Domain AggRegation Network (DARN), is able to effectively adjust the…
School of Computing Sciences, University of East Anglia（东英吉利大学计算科学学院）
Virtual Adversarial Training has recently seen a lot of success in semi-supervised learning, as well as unsupervised Domain Adaptation. However, so far it has been used on input samples in the pixel space, whereas we propose to apply it directly to feature vectors. We also discuss the unstable behaviour of entropy minimization and Decision-Boundary Iterative Refinement Training With a Teacher in Domain Adaptation, and suggest substitutes that achieve similar behaviour. By adding the aforementioned techniques to the state of the art model TA$^3$N, we either maintain competitive results or outpe…
Carnegie Mellon University（卡内基梅隆大学）；Bosch Center for AI（博世人工智能中心）
Test-time adaptation (TTA) refers to adapting neural networks to distribution shifts, with access to only the unlabeled test samples from the new domain at test-time. Prior TTA methods optimize over unsupervised objectives such as the entropy of model predictions in TENT [Wang et al., 2021], but it is unclear what exactly makes a good TTA loss. In this paper, we start by presenting a surprising phenomenon: if we attempt to meta-learn the best possible TTA loss over a wide class of functions, then we recover a function that is remarkably similar to (a temperature-scaled version of) the softmax-…
In this paper we initiate a study of non parametric contextual bandits under shape constraints on the mean reward function. Specifically, we study a setting where the context is one dimensional, and the mean reward function is isotonic with respect to this context. We propose a policy for this problem and show that it attains minimax rate optimal regret. Moreover, we show that the same policy enjoys automatic adaptation; that is, for subclasses of the parameter space where the true mean reward functions are also piecewise constant with $k$ pieces, this policy remains minimax rate optimal simul…
Computer Vision Center, Universitat Autònoma de Barcelona（巴塞罗那自治大学计算机视觉中心）；Nankai University（南开大学）；Huawei Kirin Solution（华为麒麟解决方案部）
In this paper, we investigate Source-free Open-partial Domain Adaptation (SF-OPDA), which addresses the situation where there exist both domain and category shifts between source and target domains. Under the SF-OPDA setting, which aims to address data privacy concerns, the model cannot access source data anymore during target adaptation. We propose a novel training scheme to learn a (n+1)-way classifier to predict the n source classes and the unknown class, where samples of only known source categories are available for training. Furthermore, for target adaptation, we simply adopt a weighted …
Domain adaptation arises as an important problem in statistical learning theory when the data-generating processes differ between training and test samples, respectively called source and target domains. Recent theoretical advances show that the success of domain adaptation algorithms heavily relies on their ability to minimize the divergence between the probability distributions of the source and target domains. However, minimizing this divergence cannot be done independently of the minimization of other key ingredients such as the source risk or the combined error of the ideal joint hypothes…
In this work, we face the problem of unsupervised domain adaptation with a novel deep learning approach which leverages on our finding that entropy minimization is induced by the optimal alignment of second order statistics between source and target domains. We formally demonstrate this hypothesis and, aiming at achieving an optimal alignment in practical cases, we adopt a more principled strategy which, differently from the current Euclidean approaches, deploys alignment along geodesics. Our pipeline can be implemented by adding to the standard classification loss (on the labeled source domai…
In machine learning, if the training data is an unbiased sample of an underlying distribution, then the learned classification function will make accurate predictions for new samples. However, if the training data is not an unbiased sample, then there will be differences between how the training data is distributed and how the test data is distributed. Standard classifiers cannot cope with changes in data distributions between training and test phases, and will not perform well. Domain adaptation and transfer learning are sub-fields within machine learning that are concerned with accounting fo…
Mutual information (MI) minimization has gained considerable interests in various machine learning tasks. However, estimating and minimizing MI in high-dimensional spaces remains a challenging problem, especially when only samples, rather than distribution forms, are accessible. Previous works mainly focus on MI lower bound approximation, which is not applicable to MI minimization problems. In this paper, we propose a novel Contrastive Log-ratio Upper Bound (CLUB) of mutual information. We provide a theoretical analysis of the properties of CLUB and its variational approximation. Based on this…
Massachusetts General Hospital（麻省总医院）；Harvard Medical School（哈佛医学院）
Unsupervised domain adaptation (UDA) between two significantly disparate domains to learn high-level semantic alignment is a crucial yet challenging task.~To this end, in this work, we propose exploiting low-level edge information to facilitate the adaptation as a precursor task, which has a small cross-domain gap, compared with semantic segmentation.~The precise contour then provides spatial information to guide the semantic adaptation. More specifically, we propose a multi-task framework to learn a contouring adaptation network along with a semantic segmentation adaptation network, which tak…
Provable Data Scaling Law for Meta Learning via Complexity Minimization
University of Tsukuba（筑波大学）；RIKEN AIP（理化学研究所人工智能研究中心）；SB Intuitions Corp.（SB Intuitions公司）；Kyoto University（京都大学）；Shiga University（滋贺大学）；Institute of Science Tokyo（东京科学研究所）
Zhejiang University（浙江大学）；Harvard Medical School（哈佛医学院）；University of California Santa Barbara（加州大学圣巴巴拉分校）
Domain Adaptation has been widely used to deal with the distribution shift in vision, language, multimedia etc. Most domain adaptation methods learn domain-invariant features with data from both domains available. However, such a strategy might be infeasible in practice when source data are unavailable due to data-privacy concerns. To address this issue, we propose a novel adaptation method via hypothesis transfer without accessing source data at adaptation stage. In order to fully use the limited target data, a semi-supervised mutual enhancement method is proposed, in which entropy minimizati…
Domain adaptation for semantic segmentation across datasets consisting of the same categories has seen several recent successes. However, a more general scenario is when the source and target datasets correspond to non-overlapping label spaces. For example, categories in segmentation datasets change vastly depending on the type of environment or application, yet share many valuable semantic relations. Existing approaches based on feature alignment or discrepancy minimization do not take such category shift into account. In this work, we present Cluster-to-Adapt (C2A), a computationally efficie…
CIOMP, Chinese Academy of Sciences（中国科学院长春光学精密机械与物理研究所）；Harvard University（哈佛大学）；National University of Singapore（新加坡国立大学）；China University of Mining and Technology（中国矿业大学）；Beth Israel Deaconess Medical Center（贝斯以色列女执事医疗中心）；Harvard Medical School（哈佛医学院）；The Hong Kong Polytechnic University（香港理工大学）
Deep learning is usually data starved, and the unsupervised domain adaptation (UDA) is developed to introduce the knowledge in the labeled source domain to the unlabeled target domain. Recently, deep self-training presents a powerful means for UDA, involving an iterative process of predicting the target domain and then taking the confident predictions as hard pseudo-labels for retraining. However, the pseudo-labels are usually unreliable, thus easily leading to deviated solutions with propagated errors. In this paper, we resort to the energy-based model and constrain the training of the unlabe…
The University of Tokyo（东京大学）；RIKEN（理化学研究所）；Université Paris Dauphine-PSL（巴黎多菲纳大学-PSL）
A default assumption in many machine learning scenarios is that the training and test samples are drawn from the same probability distribution. However, such an assumption is often violated in the real world due to non-stationarity of the environment or bias in sample selection. In this work, we consider a prevalent setting called covariate shift, where the input distribution differs between the training and test stages while the conditional distribution of the output given the input remains unchanged. Most of the existing methods for covariate shift adaptation are two-step approaches, which f…
The University of Texas at Austin（德克萨斯大学奥斯汀分校）；MIT-IBM Watson AI Lab（麻省理工学院-IBM沃森人工智能实验室）；Snap Inc.（Snap公司）
In this paper, we introduce a novel unsupervised domain adaptation technique for the task of 3D keypoint prediction from a single depth scan or image. Our key idea is to utilize the fact that predictions from different views of the same or similar objects should be consistent with each other. Such view consistency can provide effective regularization for keypoint prediction on unlabeled instances. In addition, we introduce a geometric alignment term to regularize predictions in the target domain. The resulting loss function can be effectively optimized via alternating minimization. We demonstr…
Duke University（杜克大学）；Microsoft Dynamics 365 AI Research（微软Dynamics 365人工智能研究院）；Microsoft Research（微软研究院）
Recent unsupervised approaches to domain adaptation primarily focus on minimizing the gap between the source and the target domains through refining the feature generator, in order to learn a better alignment between the two domains. This minimization can be achieved via a domain classifier to detect target-domain features that are divergent from source-domain features. However, by optimizing via such domain classification discrepancy, ambiguous target samples that are not smoothly distributed on the low-dimensional data manifold are often missed. To solve this issue, we propose a novel Contra…
Humans comprehend a natural scene at a single glance; painters and other visual artists, through their abstract representations, stressed this capacity to the limit. The performance of computer vision solutions matched that of humans in many problems of visual recognition. In this paper we address the problem of recognizing the genre (subject) in digitized paintings using Convolutional Neural Networks (CNN) as part of the more general dealing with abstract and/or artistic representation of scenes. Initially we establish the state of the art performance by training a CNN from scratch. In the ne…
Johannes Kepler University（约翰内斯·开普勒大学）；Software Competence Center Hagenberg GmbH（哈根贝格软件能力中心有限公司）
A novel approach for unsupervised domain adaptation for neural networks is proposed. It relies on metric-based regularization of the learning process. The metric-based regularization aims at domain-invariant latent feature representations by means of maximizing the similarity between domain-specific activation distributions. The proposed metric results from modifying an integral probability metric such that it becomes less translation-sensitive on a polynomial function space. The metric has an intuitive interpretation in the dual space as the sum of differences of higher order central moments …
The Ohio State University（俄亥俄州立大学）；Google Research（谷歌研究院）；Courant Institute of Mathematical Sciences（库朗数学科学研究所）
A key problem in a variety of applications is that of domain adaptation from a public source domain, for which a relatively large amount of labeled data with no privacy constraints is at one's disposal, to a private target domain, for which a private sample is available with very few or no labeled data. In regression problems with no privacy constraints on the source or target data, a discrepancy minimization algorithm based on several theoretical guarantees was shown to outperform a number of other adaptation algorithm baselines. Building on that approach, we design differentially private dis…
University of Edinburgh（爱丁堡大学）；Alan Turing Institute（艾伦·图灵研究所）；MPI for Intelligent Systems（马克斯·普朗克智能系统研究所）
Source-free domain adaptation (SFDA) aims to adapt a model trained on labelled data in a source domain to unlabelled data in a target domain without access to the source-domain data during adaptation. Existing methods for SFDA leverage entropy-minimization techniques which: (i) apply only to classification; (ii) destroy model calibration; and (iii) rely on the source model achieving a good level of feature-space class-separation in the target domain. We address these issues for a particularly pervasive type of domain shift called measurement shift which can be resolved by restoring the source …
