Exploiting Label Correlations Research Articles

Exploiting indirect linear correlation for label distribution learning

Label distribution learning represents the relevance of labels to samples using description degree, which can provide richer semantic information, thus finding wider applications. Exploiting label correlations is an effective approach to narrow down the hypothesis space of label distribution learning models. In existing works that utilize low-rank assumptions or label linear dependence to mine correlations, it is assumed that a label can be linearly expressed by other labels. However, this assumption can only be satisfied when there are linear dependency relationships between labels, thus the label correlation obtained by such methods is subject to certain distortion. To address this issue, this paper assumes that labels can be linearly represented by the same set of bases. The correlation between labels is represented by sharing common bases. Specifically, the paper employs matrix factorization to extract bases that can be used to represent all labels. And then designs a label distribution learning algorithm based on the property of sharing the same set of bases of the ground truth label distribution and predict label distribution. The effectiveness of the algorithm is verified through experimental validation. Generally speaking, the algorithm presented in this paper achieves optimal performance at 73.15% of the cases, with the best average ranking. In the two-tailed t-test, the algorithm in this paper exhibits statistical superiority compared to all comparison algorithms.

Multi-Label Feature Selection Via Adaptive Label Correlation Estimation

In multi-label learning, each instance is associated with multiple labels simultaneously. Multi-label data often have noisy, irrelevant, and redundant features of high dimensionality. Multi-label feature selection has received considerable attention as an effective means for dealing with high-dimensional multi-label data. Many multi-label feature selection methods exploit label correlations to help select features. However, finding label correlations and selecting features in existing multi-label feature selection methods are often two separate processes, the existence of noises and outliers in training data makes the label correlations exploited from label space less reliable. Therefore, the learned label correlations may mislead the feature selection process and result in the selection of less informative features. This article proposes a novel algorithm named ROAD, i.e., multi-label featuRe selectiOn via ADaptive label correlation estimation. ROAD jointly performs adaptive label correlation exploration and feature selection with alternating optimization to obtain reliable estimation of label correlations, which can more effectively reveal the intrinsic manifold structure among labels and lead to the selection of a more proper feature subset. Comprehensive experiments on several frequently used datasets validate the superiority of ROAD against the state-of-the-art multi-label feature selection algorithms.

Multi-task learning neural framework for categorizing sexism

Sexism, a form of oppression based on one’s sex, manifests itself in numerous ways and causes enormous suffering. In view of the growing number of experiences of sexism reported online, automatically classifying these recollections can aid in the battle against sexism by allowing gender studies researchers and government officials involved in policymaking to conduct more effective analyses. This paper investigates the 23-class fine-grained, multi-label classification of accounts (reports) of sexism. Moreover, we propose a knowledge-based cascaded multi-task framework for fine-grained multi-label sexism classification. We leverage several supporting tasks, including homogeneous and heterogeneous auxiliary tasks. Homogeneous tasks are set up without incurring any manual labeling cost and heterogeneous tasks are set up which have a high correlation with the accounts of sexism. Unlabeled accounts of sexism are utilized through unsupervised learning to help construct our multi-task setup. In addition, we incorporate a knowledge module within the framework to infuse external knowledge features into the learning process. Further, we investigate transfer learning that employs weakly labeled accounts of sexism and transfers the learning to the multi-label sexism classification. We also devise objective functions that exploit label correlations in the training data explicitly. Multiple proposed multi-task methods outperform the state-of-the-art multi-label sexism classification across five standard metrics.

Semantic and Correlation Disentangled Graph Convolutions for Multilabel Image Recognition.

Multilabel image recognition (MLR) aims to annotate an image with comprehensive labels and suffers from object occlusion or small object sizes within images. Although the existing works attempt to capture and exploit label correlations to tackle these issues, they predominantly rely on global statistical label correlations as prior knowledge for guiding label prediction, neglecting the unique label correlations present within each image. To overcome this limitation, we propose a semantic and correlation disentangled graph convolution (SCD-GC) method, which builds the image-specific graph and employs graph propagation to reason the labels effectively. Specifically, we introduce a semantic disentangling module to extract categorywise semantic features as graph nodes and develop a correlation disentangling module to extract image-specific label correlations as graph edges. Performing graph convolutions on this image-specific graph allows for better mining of difficult labels with weak visual representations. Visualization experiments reveal that our approach successfully disentangles the dominant label correlations existing within the input image. Through extensive experimentation, we demonstrate that our method achieves superior results on the challenging Microsoft COCO (MS-COCO), PASCAL visual object classes (PASCAL-VOC), NUS web image dataset (NUS-WIDE), and Visual Genome 500 (VG-500) datasets. Code is available at GitHub: https://github.com/caigitrepo/SCDGC.

Label distribution feature selection with feature weights fusion and local label correlations

Label distribution learning, where each sample is associated with a distribution of description degree, suffers from the curse of dimensionality like other traditional learning paradigms. Feature selection as a pre-processing technique is commonly used to reduce the dimension of data. Currently, label distribution feature selection focuses on exploiting label correlations under the assumption that all samples share the same label correlations. However, the corresponding label correlations for different groups of samples may tend to be different in real-world tasks. To tackle the issue, this paper presents a novel approach named label distribution feature selection with feature weights fusion and local label correlations. First, instances are separated into several clusters regarded as local samples. Then, a two-strategy approach is presented based on the local samples. For the first, the feature weights obtained from all local samples are fused into uniform feature significance, which can effectively improve the feature discrimination. For the second, to reflect the influence of label correlations locally, a local correlation matrix is encoded via the label space of local samples as additional features. Subsequently, the feature weights for this strategy are generated from a new feature space composed of additional and original features. Finally, a subset of optimal and relevant features is selected from the top-ranked features based on feature weights obtained from the above two strategies. Extensive experiments on fifteen benchmark datasets verify that the proposed algorithm is superior to five state-of-the-art approaches in terms of six evaluation metrics.

Label correlations variation for robust multi-label feature selection

Numerous high-dimension multi-label data are produced, leading to the imperative need to design excellent multi-label feature selection methods. It is of paramount importance to exploit label correlations in previous methods. However, there exist two unsolved issues in most of existing methods. First, most of existing methods explore label correlations based on the original label space with redundant and irrelevant label information. Second, previous methods either consider second-order label correlations or high-order label correlations, in fact, both two types of label correlations are significant and complementary for capturing label information. To this end, this paper proposes a robust multi-label feature selection method with both two types of label correlations. To start with, we introduce the self-expression model to consider the high-order label correlations, additionally, the l2,1-norm is imposed onto the self-expression coefficient matrix to eliminate redundant and noisy information. Furthermore, we employ a label-level regularizer to achieve pairwise label correlations. Finally, an optimization scheme with convergence proof is designed to deal with the objective function. Multiple experimental analysis results on fourteen multi-label data sets manifest that the classification performance of the proposed method outperforms other baselines.

Views Meet Labels: Personalized Relation Refinement Network for Multiview Multilabel Learning

Multiview multilabel (MVML) learning deals with objects with diverse feature vectors and rich semantics. Existing methods are built on a shared latent space among multiple views. However, they do not well capture semantic consistency and view interactions. What’s more, they neglect different contributions of each view to multilabel learning. To address these issues, a novel Personalized RElation rEfinement Network (PREEN) is proposed to sufficiently exploit associated relationships. First, common and specific information is learned in an adversarial way to prevent them from interfering with each other. Then, we adapt the standard transformer to capture cross-view interactions. Similarly, we design label-specific transformers to model label-view dependence, which associates each label with relevant views separately. Eventually, we develop an interlabel attention mechanism to exploit label correlations and refine complementary information from other labels dynamically. Extensive experiments on one real and seven public MVML datasets validate the effectiveness of our proposed PREEN.

Incorporating multiple cluster centers for multi-label learning

Multi-label learning deals with the problem that each instance is associated with multiple labels simultaneously. Most of the existing approaches aim to improve the performance of multi-label learning by exploiting label correlations. Although the data augmentation technique is widely used in many machine learning tasks, it is still unclear whether data augmentation is helpful to multi-label learning. In this article, we propose to leverage the data augmentation technique to improve the performance of multi-label learning. Specifically, we first propose a novel data augmentation approach that performs clustering on the real examples and treats the cluster centers as virtual examples, and these virtual examples naturally embody the local label correlations and label importances. Then, motivated by the cluster assumption that examples in the same cluster should have the same label, we propose a novel regularization term to bridge the gap between the real examples and virtual examples, which can promote the local smoothness of the learning function. Extensive experimental results on a number of real-world multi-label datasets clearly demonstrate that our proposed approach outperforms the state-of-the-art counterparts.

BGNN-XML: Bilateral Graph Neural Networks for Extreme Multi-label Text Classification

Extreme multi-label text classification (XMTC) aims to tag a text instance with the most relevant subset of labels from an extremely large label set. XMTC has attracted much recent attention due to massive label sets yielded by modern applications, such as news annotation and product recommendation. The main challenges of XMTC are the data <i>scalability</i> and <i>sparsity</i>, thereby leading to two issues: i) the intractability to scale to the extreme label setting, ii) the presence of long-tailed label distribution, implying that a large fraction of labels have few positive training instances. To overcome these problems, we propose BGNN-XML, a scalable graph neural network framework tailored for XMTC problems. Specifically, we exploit label correlations via excavating their co-occurrence patterns and build a label graph based on the correlation matrix. We then conduct the attributed graph clustering by performing graph convolution with a low-pass graph filter to jointly model label dependencies and label features, which induces semantic label clusters. We further propose a bilateral-branch graph isomorphism network to decouple representation learning and classifier learning for better modeling <i>tail labels</i>. Experimental results on multiple benchmark datasets demonstrate that BGNN-XML significantly outperforms state-of-the-art baselines while maintaining comparable prediction efficiency and model size.

Improving Multi-Label Learning by Correlation Embedding

In multi-label learning, each object is represented by a single instance and is associated with more than one class labels, where the labels might be correlated with each other. As we all know, exploiting label correlations can definitely improve the performance of a multi-label classification model. Existing methods mainly model label correlations in an indirect way, i.e., adding extra constraints on the coefficients or outputs of a model based on a pre-learned label correlation graph. Meanwhile, the high dimension of the feature space also poses great challenges to multi-label learning, such as high time and memory costs. To solve the above mentioned issues, in this paper, we propose a new approach for Multi-Label Learning by Correlation Embedding, namely MLLCE, where the feature space dimension reduction and the multi-label classification are integrated into a unified framework. Specifically, we project the original high-dimensional feature space to a low-dimensional latent space by a mapping matrix. To model label correlation, we learn an embedding matrix from the pre-defined label correlation graph by graph embedding. Then, we construct a multi-label classifier from the low-dimensional latent feature space to the label space, where the embedding matrix is utilized as the model coefficients. Finally, we extend the proposed method MLLCE to the nonlinear version, i.e., NL-MLLCE. The comparison experiment with the state-of-the-art approaches shows that the proposed method MLLCE has a competitive performance in multi-label learning.

Multi-Label Learning by Exploiting Label Correlations Locally

It is well known that exploiting label correlations is important for multi-label learning. Existing approaches typically exploit label correlations globally, by assuming that the label correlations are shared by all the instances. In real-world tasks, however, different instances may share different label correlations, and few correlations are globally applicable. In this paper, we propose the ML-LOC approach which allows label correlations to be exploited locally. To encode the local influence of label correlations, we derive a LOC code to enhance the feature representation of each instance. The global discrimination fitting and local correlation sensitivity are incorporated into a unified framework, and an alternating solution is developed for the optimization. Experimental results on a number of image, text and gene data sets validate the effectiveness of our approach.

Hierarchical multilabel classification by exploiting label correlations

Hierarchical multilabel classification by exploiting label correlations

Multi-label learning with missing and completely unobserved labels

Multi-label learning deals with data examples which are associated with multiple class labels simultaneously. Despite the success of existing approaches to multi-label learning, there is still a problem neglected by researchers, i.e., not only are some of the values of observed labels missing, but also some of the labels are completely unobserved for the training data. We refer to the problem as multi-label learning with missing and completely unobserved labels, and argue that it is necessary to discover these completely unobserved labels in order to mine useful knowledge and make a deeper understanding of what is behind the data. In this paper, we propose a new approach named MCUL to solve multi-label learning with Missing and Completely Unobserved Labels. We try to discover the unobserved labels of a multi-label data set with a clustering based regularization term and describe the semantic meanings of them based on the label-specific features learned by MCUL, and overcome the problem of missing labels by exploiting label correlations. The proposed method MCUL can predict both the observed and newly discovered labels simultaneously for unseen data examples. Experimental results validated over ten benchmark datasets demonstrate that the proposed method can outperform other state-of-the-art approaches on observed labels and obtain an acceptable performance on the new discovered labels as well.

Global-Local Label Correlation for Partial Multi-Label Learning

Partial Multi-label Learning (PML) addresses the scenario where each instance is assigned with multiple candidate labels, while only a subset of the labels are relevant. This task is very challenging because the training procedure can be misguided by the noisy (irrelevant) labels. Exploiting label correlations is useful for partial multi-label learning. However, the existing PML methods often ignore to explicitly and sufficiently leverage the label correlation information for handling the noisy labels. To this end, in this paper, we propose a novel <i>Global-Local Label Correlation (GLC) approach for partial multi-label learning</i>. On one hand, we introduce a label coefficient matrix to explicitly exploit the global structure information of labels from multiple subspaces. On the other hand, we present a new label manifold regularizer to capture the local label correlations to further improve the performance of our method. By jointly taking advantage of the global and local label correlations, our proposed approach achieves superior performance on both the synthetic and real-world data sets from diverse domains.

Partial multi-label learning with mutual teaching

Partial Multi-label Learning (PML) tackles the problem where each training instance is associated with a set of candidate labels that include both the relevant ground-truth labels and irrelevant false positive labels. Most of the existing PML methods try to iteratively update the confidence of each candidate label, while the estimated label confidence may be not reliable due to the cumulative error induced in the confidence updating process, especially when false positive labels dominate. In this paper, we propose a simple yet effective model called PML-MT (Partial Multi-label Learning with Mutual Teaching), in which a couple of prediction networks as well as the corresponding teacher networks are adopted to learn collaboratively and teach each other throughout the training process. Specially, the proposed PML-MT model iteratively refines the label confidence matrix through a couple of self-ensemble teacher networks and trains two prediction networks simultaneously in a mutual teaching manner. Moreover, we propose a novel regularization term to further exploit label correlations from the outputs of the prediction networks under the supervision of the refined label confidence matrix. In addition, a co-regularization term is introduced to maximize the agreement on the outputs of the couple prediction networks, so that the predictions of each network would be more reliable. Extensive experiments on synthesized and real-world PML datasets demonstrate that the proposed approach outperforms the state-of-the-art counterparts.

Regularized Matrix Factorization for Multilabel Learning With Missing Labels

This article tackles the problem of multilabel learning with missing labels. For this problem, it is widely accepted that label correlations can be used to recover the ground-truth label matrix. Most of the existing approaches impose the low-rank assumption on the observed label matrix to exploit label correlations by decomposing it into two matrices, which describe the latent factors of instances and labels, respectively. The quality of these latent factors highly influences the recovery of ground-truth labels and the construction of the multilabel classification model. In this article, we propose recovering the ground-truth label matrix by regularized matrix factorization. Specifically, the latent factors of instances are regularized by the local topological structure derived from the feature space, which can be further used to induce an effective multilabel model. Moreover, the latent factors of labels and the label correlations are mutually adapted via label manifold regularization. In this way, the recovery of the ground-truth label matrix and the construction of the multilabel classification model are optimized jointly and can benefit from the regularized matrix factorization. Extensive experimental studies show that the proposed approach significantly outperforms the state-of-the-art algorithms on both full-label and missing-label data.

Multilabel Text Classification With Incomplete Labels: A Safe Generative Model With Label Manifold Regularization and Confidence Constraint

In multilabel text classification, the label information of training instances may be incomplete due to various factors, such as expensively manual labeling, errors, etc. Hence, imposing big challenges to training classifiers. To tackle the problem of the incomplete labels, we proposed to exploit label correlations, and further jointly leverage manifold regularization and label confidence constraint. On one hand, the label manifold regularization preserves the local label structure to handle the high-level incomplete labels. On the other hand, the label confidence constraint, following the likelihood theory, avoids overestimating the weights of negative labels, leading to a safer inference. Upon the ideas, we propose a novel generative model, namely B ayesian M odel with L abel manifold R egularization and label C onfidence constraint (BM-LRC), solved by generalized expectation maximization. Empirical results on popular benchmark data sets show that BM-LRC achieves competitive performance, especially when the available labels are scarce.

Label Embedding for Multi-label Classification Via Dependence Maximization

Multi-label classification has aroused extensive attention in various fields. With the emergence of high-dimensional label space, academia has devoted to performing label embedding in recent years. Whereas current embedding approaches do not take feature space correlation sufficiently into consideration or require an encoding function while learning embedded space. Besides, few of them can be spread to track the missing labels. In this paper, we propose a Label Embedding method via Dependence Maximization (LEDM), which obtains the latent space on which the label and feature information can be embedded simultaneously. To end this, the low-rank factorization model on the label matrix is applied to exploit label correlations instead of the encoding process. The dependence between feature space and label space is increased by the Hilbert–Schmidt independence criterion to facilitate the predictability. The proposed LEDM can be easily extended the missing labels in learning embedded space at the same time. Comprehensive experimental results on data sets validate the effectiveness of our approach over the state-of-art methods on both complete-label and missing-label cases.

Weakly-supervised multi-label learning with noisy features and incomplete labels

Weakly-supervised multi-label learning has emerged as a hot topic more recently. Most existing methods deal with such problem by learning from the data where the label assignments are incomplete while the feature information is ideal. However, in many real applications, due to the influence of occlusion, illumination and low-resolution, the acquired features are often noisy, which may reduce the robustness of the learning model. In this paper, to overcome the above shortcoming, we propose a novel weakly-supervised multi-label learning framework called WML-LSC, where the low-rank and sparse constrain schemes are jointly incorporated to capture the desired feature information. Specifically, we first decompose the observed feature matrix into an ideal feature matrix and an outlier matrix. Considering that similar instances usually share similar visual characteristics, we constrain the ideal feature matrix to be low-rank. Meanwhile, a reasonable assumption is that the noise is sparse compared with the feature matrix, which leads outlier matrix to be sparse. In addition, a linear self-recovery model is adopted to reconstruct the incomplete label assignment matrix by exploiting label correlations. Finally, the desired model is trained on the ideal feature matrix and the refined label matrix. Extensive experimental results demonstrate that our proposed method can achieve superior and comparable performance against state-of-the-art methods.

Disentangling, Embedding and Ranking Label Cues for Multi-Label Image Recognition

Multi-label image recognition is a fundamental but challenging computer vision and multimedia task. Great progress has been achieved by exploiting label correlations among these multiple labels associated with a single image, which is the most crucial issue for multi-label image recognition. In this paper, to explicitly model label correlations, we propose a unified deep learning framework to Disentangle, Embed and Rank (DER) the corresponding label cues. Specifically, we first obtain class-aware disentangled maps (CADMs) by reforming deep activations in accordance with the class-specific recognition weights. Then, after transforming CADMs into the corresponding label vectors, we propose an embedding operation from a metric learning perspective to pull the relevant label vectors together and push irrelevant label vectors away. Furthermore, a ranking operation is employed, which aims to accurately and robustly measure the similarity/dissimilarity of these label vectors. Our model can be trained in an end-to-end manner with only image-level supervision, during which the proposed embedding and ranking operations can contribute to the CADMs learning through back-propagation. In addition, the obtained CADMs are aggregated and further used as an essential feature stream for the final multi-label classification. We conduct extensive experiments on three commonly used multi-label benchmark datasets. Quantitative results show that our model can significantly and consistently outperform previous competitive methods. Moreover, qualitative analysis of our DER proposal also reveals the effectiveness of our proposed model.