When Invariant Representation Learning Meets Label Shift: Insufficiency and Theoretical Insights.

As a crucial step toward real-world learning scenarios with changing environments, dataset shift theory and invariant representation learning algorithm have been extensively studied to relax the identical distribution assumption in classical learning setting. Among the different assumptions on the essential of shifting distributions, generalized label shift (GLS) is the latest developed one which shows great potential to deal with the complex factors within the shift. In this paper, we aim to explore the limitations of current dataset shift theory and algorithm, and further provide new insights by presenting a comprehensive understanding of GLS. From theoretical aspect, two informative generalization bounds are derived, and the GLS learner are proved to be sufficiently close to optimal target model from the Bayesian perspective. The main results show the insufficiency of invariant representation learning, and prove the sufficiency and necessity of GLS correction for generalization, which provide theoretical supports and innovations for exploring generalizable model under dataset shift. From methodological aspect, we provide a unified view of existing shift correction frameworks, and propose a kernel embedding-based correction algorithm (KECA) to minimize the generalization error and achieve successful knowledge transfer. Both theoretical results and extensive experiment evaluations demonstrate the sufficiency and necessity of GLS correction for addressing dataset shift and the superiority of proposed algorithm.

Geometric Understanding of Discriminability and Transferability for Visual Domain Adaptation.

To overcome the restriction of identical distribution assumption, invariant representation learning for unsupervised domain adaptation (UDA) has made significant advances in computer vision and pattern recognition communities. In UDA scenario, the training and test data belong to different domains while the task model is learned to be invariant. Recently, empirical connections between transferability and discriminability have received increasing attention, which is the key to understand the invariant representations. However, theoretical study of these abilities and in-depth analysis of the learned feature structures are unexplored yet. In this work, we systematically analyze the essentials of transferability and discriminability from the geometric perspective. Our theoretical results provide insights into understanding the co-regularization relation and prove the possibility of learning these abilities. From methodology aspect, the abilities are formulated as geometric properties between domain/cluster subspaces (i.e., orthogonality and equivalence) and characterized as the relation between the norms/ranks of multiple matrices. Two optimization-friendly learning principles are derived, which also ensure some intuitive explanations. Moreover, a feasible range for the co-regularization parameters is deduced to balance the learning of geometric structures. Based on the theoretical results, a geometry-oriented model is proposed for enhancing the transferability and discriminability via nuclear norm optimization. Extensive experiment results validate the effectiveness of the proposed model in empirical applications, and verify that the geometric abilities can be sufficiently learned in the derived feasible range.

Cross-site prognosis prediction for nasopharyngeal carcinoma from incomplete multi-modal data.

Accurate prognosis prediction for nasopharyngeal carcinoma based on magnetic resonance (MR) images assists in the guidance of treatment intensity, thus reducing the risk of recurrence and death. To reduce repeated labor and sufficiently explore domain knowledge, aggregating labeled/annotated data from external sites enables us to train an intelligent model for a clinical site with unlabeled data. However, this task suffers from the challenges of incomplete multi-modal examination data fusion and image data heterogeneity among sites. This paper proposes a cross-site survival analysis method for prognosis prediction of nasopharyngeal carcinoma from domain adaptation viewpoint. Utilizing a Cox model as the basic framework, our method equips it with a cross-attention based multi-modal fusion regularization. This regularization model effectively fuses the multi-modal information from multi-parametric MR images and clinical features onto a domain-adaptive space, despite the absence of some modalities. To enhance the feature discrimination, we also extend the contrastive learning technique to censored data cases. Compared with the conventional approaches which directly deploy a trained survival model in a new site, our method achieves superior prognosis prediction performance in cross-site validation experiments. These results highlight the key role of cross-site adaptability of our method and support its value in clinical practice.

BuresNet: Conditional Bures Metric for Transferable Representation Learning.

As a fundamental manner for learning and cognition, transfer learning has attracted widespread attention in recent years. Typical transfer learning tasks include unsupervised domain adaptation (UDA) and few-shot learning (FSL), which both attempt to sufficiently transfer discriminative knowledge from the training environment to the test environment to improve the model's generalization performance. Previous transfer learning methods usually ignore the potential conditional distribution shift between environments. This leads to the discriminability degradation in the test environments. Therefore, how to construct a learnable and interpretable metric to measure and then reduce the gap between conditional distributions is very important in the literature. In this article, we design the Conditional Kernel Bures (CKB) metric for characterizing conditional distribution discrepancy, and derive an empirical estimation with convergence guarantee. CKB provides a statistical and interpretable approach, under the optimal transportation framework, to understand the knowledge transfer mechanism. It is essentially an extension of optimal transportation from the marginal distributions to the conditional distributions. CKB can be used as a plug-and-play module and placed onto the loss layer in deep networks, thus, it plays the bottleneck role in representation learning. From this perspective, the new method with network architecture is abbreviated as BuresNet, and it can be used extract conditional invariant features for both UDA and FSL tasks. BuresNet can be trained in an end-to-end manner. Extensive experiment results on several benchmark datasets validate the effectiveness of BuresNet.

Multi-Source Unsupervised Domain Adaptation via Pseudo Target Domain.

Multi-source domain adaptation (MDA) aims to transfer knowledge from multiple source domains to an unlabeled target domain. MDA is a challenging task due to the severe domain shift, which not only exists between target and source but also exists among diverse sources. Prior studies on MDA either estimate a mixed distribution of source domains or combine multiple single-source models, but few of them delve into the relevant information among diverse source domains. For this reason, we propose a novel MDA approach, termed Pseudo Target for MDA (PTMDA). Specifically, PTMDA maps each group of source and target domains into a group-specific subspace using adversarial learning with a metric constraint, and constructs a series of pseudo target domains correspondingly. Then we align the remainder source domains with the pseudo target domain in the subspace efficiently, which allows to exploit additional structured source information through the training on pseudo target domain and improves the performance on the real target domain. Besides, to improve the transferability of deep neural networks (DNNs), we replace the traditional batch normalization layer with an effective matching normalization layer, which enforces alignments in latent layers of DNNs and thus gains further promotion. We give theoretical analysis showing that PTMDA as a whole can reduce the target error bound and leads to a better approximation of the target risk in MDA settings. Extensive experiments demonstrate PTMDA's effectiveness on MDA tasks, as it outperforms state-of-the-art methods in most experimental settings.

Cross-Site Severity Assessment of COVID-19 From CT Images via Domain Adaptation.

Early and accurate severity assessment of Coronavirus disease 2019 (COVID-19) based on computed tomography (CT) images offers a great help to the estimation of intensive care unit event and the clinical decision of treatment planning. To augment the labeled data and improve the generalization ability of the classification model, it is necessary to aggregate data from multiple sites. This task faces several challenges including class imbalance between mild and severe infections, domain distribution discrepancy between sites, and presence of heterogeneous features. In this paper, we propose a novel domain adaptation (DA) method with two components to address these problems. The first component is a stochastic class-balanced boosting sampling strategy that overcomes the imbalanced learning problem and improves the classification performance on poorly-predicted classes. The second component is a representation learning that guarantees three properties: 1) domain-transferability by prototype triplet loss, 2) discriminant by conditional maximum mean discrepancy loss, and 3) completeness by multi-view reconstruction loss. Particularly, we propose a domain translator and align the heterogeneous data to the estimated class prototypes (i.e., class centers) in a hyper-sphere manifold. Experiments on cross-site severity assessment of COVID-19 from CT images show that the proposed method can effectively tackle the imbalanced learning problem and outperform recent DA approaches.

Unsupervised Domain Adaptation via Discriminative Manifold Propagation.

Unsupervised domain adaptation is effective in leveraging rich information from a labeled source domain to an unlabeled target domain. Though deep learning and adversarial strategy made a significant breakthrough in the adaptability of features, there are two issues to be further studied. First, hard-assigned pseudo labels on the target domain are arbitrary and error-prone, and direct application of them may destroy the intrinsic data structure. Second, batch-wise training of deep learning limits the characterization of the global structure. In this paper, a Riemannian manifold learning framework is proposed to achieve transferability and discriminability simultaneously. For the first issue, this framework establishes a probabilistic discriminant criterion on the target domain via soft labels. Based on pre-built prototypes, this criterion is extended to a global approximation scheme for the second issue. Manifold metric alignment is adopted to be compatible with the embedding space. The theoretical error bounds of different alignment metrics are derived for constructive guidance. The proposed method can be used to tackle a series of variants of domain adaptation problems, including both vanilla and partial settings. Extensive experiments have been conducted to investigate the method and a comparative study shows the superiority of the discriminative manifold learning framework.

Hyperspectral Image Classification via Discriminant Gabor Ensemble Filter.

For a broad range of applications, hyperspectral image (HSI) classification is a hot topic in remote sensing, and convolutional neural network (CNN)-based methods are drawing increasing attention. However, to train millions of parameters in CNN requires a large number of labeled training samples, which are difficult to collect. A conventional Gabor filter can effectively extract spatial information with different scales and orientations without training, but it may be missing some important discriminative information. In this article, we propose the Gabor ensemble filter (GEF), a new convolutional filter to extract deep features for HSI with fewer trainable parameters. GEF filters each input channel by some fixed Gabor filters and learnable filters simultaneously, then reduces the dimensions by some learnable 1×1 filters to generate the output channels. The fixed Gabor filters can extract common features with different scales and orientations, while the learnable filters can learn some complementary features that Gabor filters cannot extract. Based on GEF, we design a network architecture for HSI classification, which extracts deep features and can learn from limited training samples. In order to simultaneously learn more discriminative features and an end-to-end system, we propose to introduce the local discriminant structure for cross-entropy loss by combining the triplet hard loss. Results of experiments on three HSI datasets show that the proposed method has significantly higher classification accuracy than other state-of-the-art methods. Moreover, the proposed method is speedy for both training and testing.

Learning representation for multiple biological networks via a robust graph regularized integration approach.

Learning node representation is a fundamental problem in biological network analysis, as compact representation features reveal complicated network structures and carry useful information for downstream tasks such as link prediction and node classification. Recently, multiple networks that profile objects from different aspects are increasingly accumulated, providing the opportunity to learn objects from multiple perspectives. However, the complex common and specific information across different networks pose challenges to node representation methods. Moreover, ubiquitous noise in networks calls for more robust representation. To deal with these problems, we present a representation learning method for multiple biological networks. First, we accommodate the noise and spurious edges in networks using denoised diffusion, providing robust connectivity structures for the subsequent representation learning. Then, we introduce a graph regularized integration model to combine refined networks and compute common representation features. By using the regularized decomposition technique, the proposed model can effectively preserve the common structural property of different networks and simultaneously accommodate their specific information, leading to a consistent representation. A simulation study shows the superiority of the proposed method on different levels of noisy networks. Three network-based inference tasks, including drug-target interaction prediction, gene function identification and fine-grained species categorization, are conducted using representation features learned from our method. Biological networks at different scales and levels of sparsity are involved. Experimental results on real-world data show that the proposed method has robust performance compared with alternatives. Overall, by eliminating noise and integrating effectively, the proposed method is able to learn useful representations from multiple biological networks.

Learning Target-Domain-Specific Classifier for Partial Domain Adaptation.

Unsupervised domain adaptation (UDA) aims at reducing the distribution discrepancy when transferring knowledge from a labeled source domain to an unlabeled target domain. Previous UDA methods assume that the source and target domains share an identical label space, which is unrealistic in practice since the label information of the target domain is agnostic. This article focuses on a more realistic UDA scenario, i.e., partial domain adaptation (PDA), where the target label space is subsumed to the source label space. In the PDA scenario, the source outliers that are absent in the target domain may be wrongly matched to the target domain (technically named negative transfer), leading to performance degradation of UDA methods. This article proposes a novel target-domain-specific classifier learning-based domain adaptation (TSCDA) method. TSCDA presents a soft-weighed maximum mean discrepancy criterion to partially align feature distributions and alleviate negative transfer. Also, it learns a target-specific classifier for the target domain with pseudolabels and multiple auxiliary classifiers to further address the classifier shift. A module named peers-assisted learning is used to minimize the prediction difference between multiple target-specific classifiers, which makes the classifiers more discriminant for the target domain. Extensive experiments conducted on three PDA benchmark data sets show that TSCDA outperforms other state-of-the-art methods with a large margin, e.g., 4% and 5.6% averagely on Office-31 and Office-Home, respectively.

Heterogeneous Domain Adaptation via Covariance Structured Feature Translators.

Domain adaptation (DA) and transfer learning with statistical property description is very important in image analysis and data classification. This article studies the domain adaptive feature representation problem for the heterogeneous data, of which both the feature dimensions and the sample distributions across domains are so different that their features cannot be matched directly. To transfer the discriminant information efficiently from the source domain to the target domain, and then enhance the classification performance for the target data, we first introduce two projection matrices specified for different domains to transform the heterogeneous features into a shared space. We then propose a joint kernel regression model to learn the regression variable, which is called feature translator in this article. The novelty focuses on the exploration of optimal experimental design (OED) to deal with the heterogeneous and nonlinear DA by seeking the covariance structured feature translators (CSFTs). An approximate and efficient method is proposed to compute the optimal data projections. Comprehensive experiments are conducted to validate the effectiveness and efficacy of the proposed model. The results show the state-of-the-art performance of our method in heterogeneous DA.

Learning Kernel for Conditional Moment-Matching Discrepancy-Based Image Classification.

Conditional maximum mean discrepancy (CMMD) can capture the discrepancy between conditional distributions by drawing support from nonlinear kernel functions; thus, it has been successfully used for pattern classification. However, CMMD does not work well on complex distributions, especially when the kernel function fails to correctly characterize the difference between intraclass similarity and interclass similarity. In this paper, a new kernel learning method is proposed to improve the discrimination performance of CMMD. It can be operated with deep network features iteratively and thus denoted as KLN for abbreviation. The CMMD loss and an autoencoder (AE) are used to learn an injective function. By considering the compound kernel, that is, the injective function with a characteristic kernel, the effectiveness of CMMD for data category description is enhanced. KLN can simultaneously learn a more expressive kernel and label prediction distribution; thus, it can be used to improve the classification performance in both supervised and semisupervised learning scenarios. In particular, the kernel-based similarities are iteratively learned on the deep network features, and the algorithm can be implemented in an end-to-end manner. Extensive experiments are conducted on four benchmark datasets, including MNIST, SVHN, CIFAR-10, and CIFAR-100. The results indicate that KLN achieves the state-of-the-art classification performance.

Dual Adversarial Autoencoders for Clustering.

As a powerful approach for exploratory data analysis, unsupervised clustering is a fundamental task in computer vision and pattern recognition. Many clustering algorithms have been developed, but most of them perform unsatisfactorily on the data with complex structures. Recently, adversarial autoencoder (AE) (AAE) shows effectiveness on tackling such data by combining AE and adversarial training, but it cannot effectively extract classification information from the unlabeled data. In this brief, we propose dual AAE (Dual-AAE) which simultaneously maximizes the likelihood function and mutual information between observed examples and a subset of latent variables. By performing variational inference on the objective function of Dual-AAE, we derive a new reconstruction loss which can be optimized by training a pair of AEs. Moreover, to avoid mode collapse, we introduce the clustering regularization term for the category variable. Experiments on four benchmarks show that Dual-AAE achieves superior performance over state-of-the-art clustering methods. In addition, by adding a reject option, the clustering accuracy of Dual-AAE can reach that of supervised CNN algorithms. Dual-AAE can also be used for disentangling style and content of images without using supervised information.

Generalized Conditional Domain Adaptation: A Causal Perspective With Low-Rank Translators.

Learning domain adaptive features aims to enhance the classification performance of the target domain by exploring the discriminant information from an auxiliary source set. Let X denote the feature and Y as the label. The most typical problem to be addressed is that P XY has a so large variation between different domains that classification in the target domain is difficult. In this paper, we study the generalized conditional domain adaptation (DA) problem, in which both P Y and P X|Y change across domains, in a causal perspective. We propose transforming the class conditional probability matching to the marginal probability matching problem, under a proper assumption. We build an intermediate domain by employing a regression model. In order to enforce the most relevant data to reconstruct the intermediate representations, a low-rank constraint is placed on the regression model for regularization. The low-rank constraint underlines a global algebraic structure between different domains, and stresses the group compactness in representing the samples. The new model is considered under the discriminant subspace framework, which is favorable in simultaneously extracting the classification information from the source domain and adaptation information across domains. The model can be solved by an alternative optimization manner of quadratic programming and the alternative Lagrange multiplier method. To the best of our knowledge, this paper is the first to exploit low-rank representation, from the source domain to the intermediate domain, to learn the domain adaptive features. Comprehensive experimental results validate that the proposed method provides better classification accuracies with DA, compared with well-established baselines.

Discriminative Residual Analysis for Image Set Classification with Posture and Age Variations.

Image set recognition has been widely applied in many practical problems like real-time video retrieval and image caption tasks. Due to its superior performance, it has grown into a significant topic in recent years. However, images with complicated variations, e.g., postures and human ages, are difficult to address, as these variations are continuous and gradual with respect to image appearance. Consequently, the crucial point of image set recognition is to mine the intrinsic connection or structural information from the image batches with variations. In this work, a Discriminant Residual Analysis (DRA) method is proposed to improve the classification performance by discovering discriminant features in related and unrelated groups. Specifically, DRA attempts to obtain a powerful projection which casts the residual representations into a discriminant subspace. Such a projection subspace is expected to magnify the useful information of the input space as much as possible, then the relation between the training set and the test set described by the given metric or distance will be more precise in the discriminant subspace. We also propose a nonfeasance strategy by defining another approach to construct the unrelated groups, which help to reduce furthermore the cost of sampling errors. Two regularization approaches are used to deal with the probable small sample size problem. Extensive experiments are conducted on benchmark databases, and the results show superiority and efficiency of the new methods.

Radial Basis Functions With Adaptive Input and Composite Trend Representation for Portfolio Selection.

We propose a set of novel radial basis functions with adaptive input and composite trend representation (AICTR) for portfolio selection (PS). Trend representation of asset price is one of the main information to be exploited in PS. However, most state-of-the-art trend representation-based systems exploit only one kind of trend information and lack effective mechanisms to construct a composite trend representation. The proposed system exploits a set of RBFs with multiple trend representations, which improves the effectiveness and robustness in price prediction. Moreover, the input of the RBFs automatically switches to the best trend representation according to the recent investing performance of different price predictions. We also propose a novel objective to combine these RBFs and select the portfolio. Extensive experiments on six benchmark data sets (including a new challenging data set that we propose) from different real-world stock markets indicate that the proposed RBFs effectively combine different trend representations and AICTR achieves state-of-the-art investing performance and risk control. Besides, AICTR withstands the reasonable transaction costs and runs fast; hence, it is applicable to real-world financial environments.

Kernel Embedding Multiorientation Local Pattern for Image Representation.

Local feature descriptor plays a key role in different image classification applications. Some of these methods such as local binary pattern and image gradient orientations have been proven effective to some extent. However, such traditional descriptors which only utilize single-type features, are deficient to capture the edges and orientations information and intrinsic structure information of images. In this paper, we propose a kernel embedding multiorientation local pattern (MOLP) to address this problem. For a given image, it is first transformed by gradient operators in local regions, which generate multiorientation gradient images containing edges and orientations information of different directions. Then the histogram feature which takes into account the sign component and magnitude component, is extracted to form the refined feature from each orientation gradient image. The refined feature captures more information of the intrinsic structure, and is effective for image representation and classification. Finally, the multiorientation refined features are automatically fused in the kernel embedding discriminant subspace learning model. The extensive experiments on various image classification tasks, such as face recognition, texture classification, object categorization, and palmprint recognition show that MOLP could achieve competitive performance with those state-of-the art methods.

A Peak Price Tracking-Based Learning System for Portfolio Selection.

We propose a novel linear learning system based on the peak price tracking (PPT) strategy for portfolio selection (PS). Recently, the topic of tracking control attracts intensive attention and some novel models are proposed based on backstepping methods, such that the system output tracks a desired trajectory. The proposed system has a similar evolution with a transform function that aggressively tracks the increasing power of different assets. As a result, the better performing assets will receive more investment. The proposed PPT objective can be formulated as a fast backpropagation algorithm, which is suitable for large-scale and time-limited applications, such as high-frequency trading. Extensive experiments on several benchmark data sets from diverse real financial markets show that PPT outperforms other state-of-the-art systems in computational time, cumulative wealth, and risk-adjusted metrics. It suggests that PPT is effective and even more robust than some defensive systems in PS.

Learning Kernel Extended Dictionary for Face Recognition.

A sparse representation classifier (SRC) and a kernel discriminant analysis (KDA) are two successful methods for face recognition. An SRC is good at dealing with occlusion, while a KDA does well in suppressing intraclass variations. In this paper, we propose kernel extended dictionary (KED) for face recognition, which provides an efficient way for combining KDA and SRC. We first learn several kernel principal components of occlusion variations as an occlusion model, which can represent the possible occlusion variations efficiently. Then, the occlusion model is projected by KDA to get the KED, which can be computed via the same kernel trick as new testing samples. Finally, we use structured SRC for classification, which is fast as only a small number of atoms are appended to the basic dictionary, and the feature dimension is low. We also extend KED to multikernel space to fuse different types of features at kernel level. Experiments are done on several large-scale data sets, demonstrating that not only does KED get impressive results for nonoccluded samples, but it also handles the occlusion well without overfitting, even with a single gallery sample per subject.

Enhanced Local Gradient Order Features and Discriminant Analysis for Face Recognition.

Robust descriptor-based subspace learning with complex data is an active topic in pattern analysis and machine intelligence. A few researches concentrate the optimal design on feature representation and metric learning. However, traditionally used features of single-type, e.g., image gradient orientations (IGOs), are deficient to characterize the complete variations in robust and discriminant subspace learning. Meanwhile, discontinuity in edge alignment and feature match are not been carefully treated in the literature. In this paper, local order constrained IGOs are exploited to generate robust features. As the difference-based filters explicitly consider the local contrasts within neighboring pixel points, the proposed features enhance the local textures and the order-based coding ability, thus discover intrinsic structure of facial images further. The multimodal features are automatically fused in the most discriminant subspace. The utilization of adaptive interaction function suppresses outliers in each dimension for robust similarity measurement and discriminant analysis. The sparsity-driven regression model is modified to adapt the classification issue of the compact feature representation. Extensive experiments are conducted by using some benchmark face data sets, e.g., of controlled and uncontrolled environments, to evaluate our new algorithm.