Multi-label Learning Problem Research Articles

A novel machine learning model based on sparse structure learning with adaptive graph regularization for predicting drug side effects

Drug side effects are closely related to the success and failure of drug development. Here we present a novel machine learning method for side effect prediction. The proposed method treats side effect prediction as a multi-label learning problem and uses sparse structure learning to model the relationships between side effects. Additionally, the proposed method adopts the adaptive graph regularization strategy to explore the local structure in drug data and fuse multiple types of drug features. An alternating optimization algorithm is proposed to solve the optimization problem. We collected chemical structures and biological pathway features of drugs as the inputs of our method to predict drug side effects. The results of the cross-validation experiment showed that our method could significantly improve the prediction performance compared to the other state-of-the-art methods. Besides, our model is highly interpretable. It could learn the drug neighbourhood relationships, side effect relationships, and drug features related to side effects. We systematically validated the information extracted by the model with independent data. Some prediction results could also be supported by literature reports. The proposed method could be applied to integrate both chemical and biological data to predict side effects and helps improve drug safety.

Multi-Label Attribute Selection of Arrhythmia for Electrocardiogram Signals with Fusion Learning.

There are three primary challenges in the automatic diagnosis of arrhythmias by electrocardiogram (ECG): the significant variation among individual patients, the multiple pathologies in the ECG signal and the high cost in annotating clinical ECG with the corresponding labels. Traditional ECG processing approaches rely heavily on prior knowledge, such as those from feature extraction and waveform analysis. The preprocessing for prior knowledge incurs computational overhead. Furthermore, standard deep learning methods do not fully consider the dynamic temporal, spatial and multi-labeling characteristics of ECG data. In clinical ECG waveforms, it is common to see multi-labeling in which a patient is labeled with multiple classes of arrhythmias. However, multiclass approaches in current research mainly solve the multi-label machine learning problem, ignoring the correlation between diseases, resulting in information loss. In this paper, an arrhythmia detection and classification scheme called multi-label fusion deep learning is proposed. The objective is to build a unified system with automatic feature learning which supports effective multi-label classification. First, a multi-label ECG-based feature selection method is combined with a matrix decomposition and sparse learning theory. The optimal feature subset is selected as a preprocessing algorithm for ECG data. A multi-label classifier is then constructed by fusing CNN and RNN networks to fully exploit the interactions and features of the time and space dimensions. The experimental result demonstrates that the proposed method can achieve a state-of-the-art performance compared to other algorithms in multi-label database experiments.

A multi-label learning framework for predicting antibiotic resistance genes via dual-view modeling.

The increasing prevalence of antibiotic resistance has become a global health crisis. For the purpose of safety regulation, it is of high importance to identify antibiotic resistance genes (ARGs) in bacteria. Although culture-based methods can identify ARGs relatively more accurately, the identifying process is time-consuming and specialized knowledge is required. With the rapid development of whole genome sequencing technology, researchers attempt to identify ARGs by computing sequence similarity from public databases. However, these computational methods might fail to detect ARGs due to the low sequence identity to known ARGs. Moreover, existing methods cannot effectively address the issue of multidrug resistance prediction for ARGs, which is a great challenge to clinical treatments. To address the challenges, we propose an end-to-end multi-label learning framework for predicting ARGs. More specifically, the task of ARGs prediction is modeled as a problem of multi-label learning, and a deep neural network-based end-to-end framework is proposed, in which a specific loss function is introduced to employ the advantage of multi-label learning for ARGs prediction. In addition, a dual-view modeling mechanism is employed to make full use of the semantic associations among two views of ARGs, i.e. sequence-based information and structure-based information. Extensive experiments are conducted on publicly available data, and experimental results demonstrate the effectiveness of the proposed framework on the task of ARGs prediction.

Weak multi-label learning with missing labels via instance granular discrimination

In multi-label learning, each training instance is associated with multiple class labels. It is typical in reality that relevant labels are partially missing and only a part of labels are valid, resulting in the problem of weak multi-label learning with missing labels. It is still an evident challenge to estimate the ground-truth label matrix and to generate a prediction function, especially on the multi-label data with a large number of missing labels. In this paper, we propose a multi-label learning framework within which feature structure and label manifold are both utilized to recover the incomplete label matrix and to train the classification model simultaneously. To mitigate the imbalanced risks brought by the sparse label issue, a self-adaptive penalty factor is imposed on the deviated predictions of different labels. Moreover, instance granular discrimination is incorporated in the framework to characterize the approximate distribution structure of data. Matrix vectorization, cave-convex programming (CCCP), and block coordinate descent techniques are employed to solve the proposed optimization problem. Extensive experiments illustrate the superiority of the proposed method against some well-established methods.

Multi-Label Learning on Tensor Product Graph

A large family of graph-based semi-supervised algorithms have been developed intuitively and pragmatically for the multi-label learning problem. These methods, however, only implicitly exploited the label correlation, as either part of graph weight or an additional constraint, to improve overall classification performance. Despite their seemingly quite different formulations, we show that all existing approaches can be uniformly referred to as a Label Propagation (LP) or Random Walk with Restart (RWR) on a Cartesian Product Graph (CPG). Inspired by this discovery, we introduce a new framework for multi-label classification task, employing the Tensor Product Graph (TPG) — the tensor product of the data graph with the class (label) graph — in which not only the intra-class but also the inter-class associations are explicitly represented as weighted edges among graph vertices. In stead of computing directly on TPG, we derive an iterative algorithm, which is guaranteed to converge and with the same computational complexity and the same amount of storage as the standard label propagation on the original data graph. Applications to four benchmark multi-label data sets illustrate that our method outperforms several state-of-the-art approaches.

Multi-label classification with local pairwise and high-order label correlations using graph partitioning

In multi-label learning problems, the class labels are correlated and the label correlations can be leveraged to improve the predictive performance of a classifier. Methods that consider high-order correlations in the label space, usually do not utilize pairwise correlations. In most of these methods, label correlations are considered as prior knowledge, which can be misleading in problems with noisy or missing labels. In such cases, learning the label correlation as part of the model training task is more effective. In this paper, a rule-based evolutionary multi-label classification method is proposed that incorporates the local label correlations through the high-order label subsets and pairwise dependencies. Graph structures are employed to model the label dependencies and the estimated label similarities are used to obtain more accurate label sets for the classification rules. To refine the high-order label relations, a novel hierarchical density-based clustering method is proposed to obtain a k-way partitioning for the label graphs based on their pairwise correlations. The effectiveness of the proposed method is experimented on multiple benchmark datasets from different domains and compared with several well-known multi-label classification algorithms. The proposed method has shown the highest average rank along multiple metrics and the results are consistently better than or similar to the compared methods with statistical significance.

Large scale multi-label learning using Gaussian processes

We introduce a Gaussian process latent factor model for multi-label classification that can capture correlations among class labels by using a small set of latent Gaussian process functions. To address computational challenges, when the number of training instances is very large, we introduce several techniques based on variational sparse Gaussian process approximations and stochastic optimization. Specifically, we apply doubly stochastic variational inference that sub-samples data instances and classes which allows us to cope with Big Data. Furthermore, we show it is possible and beneficial to optimize over inducing points, using gradient-based methods, even in very high dimensional input spaces involving up to hundreds of thousands of dimensions. We demonstrate the usefulness of our approach on several real-world large-scale multi-label learning problems.

Classifier Chains: A Review and Perspectives

The family of methods collectively known as classifier chains has become a popular approach to multi-label learning problems. This approach involves chaining together off-the-shelf binary classifiers in a directed structure, such that individual label predictions become features for other classifiers. Such methods have proved flexible and effective and have obtained state-of-the-art empirical performance across many datasets and multi-label evaluation metrics. This performance led to further studies of the underlying mechanism and efficacy, and investigation into how it could be improved. In the recent decade, numerous studies have explored the theoretical underpinnings of classifier chains, and many improvements have been made to the training and inference procedures, such that this method remains among the best options for multi-label learning. Given this past and ongoing interest, which covers a broad range of applications and research themes, the goal of this work is to provide a review of classifier chains, a survey of the techniques and extensions provided in the literature, as well as perspectives for this approach in the domain of multi-label classification in the future. We conclude positively, with a number of recommendations for researchers and practitioners, as well as outlining key issues for future research.

Adaptive feature selection in PET scans based on shared information and multi-label learning

Feature selection for tumor treatment outcome prediction in PET scans amounts to determine the best predictors in order to classify treatment outcomes for unseen data. Several challenges have to be addressed, notably the dissensus about the most predictive radiomic features and the relatively small-sized and imbalanced training samples. To overcome these issues, we propose a multi-label learning (MLL) problem formulation that we associated to the technique of feature selection with shared information among multiple tasks. In fact, we opt to simultaneously mining the shared knowledge across similar tasks from MRI and fused PET/MRI images in order to sort PET features according to their relevancy. Then, by repeatedly training an SVM model, different feature sets are elected to be informative for each iteration. Thus, by associating these sets with their correspondents in other iterations, the MLL formulation is performed in order to adequately adapt both feature sets and training instances according to their fit with unseen data. The adaptive synthetic sampling technique is adapted to the framework of MLL data balance. After that, the global and local label correlation technique is used for multi-labeling, while taking into consideration the capacity of MLL on recommending the best sets of features and training instances. Finally, an SVM classifier is applied with both selected features and training sub-samples in order to output prediction results. Conducted experiments show that attributing each feature set to a particular set of instances and training the model with balanced subset of the whole training data allows to outperform relevant methods from the state of the art.

Multi-Label Data Fusion to Support Agricultural Vulnerability Assessments

Identifying crop species and varieties adaptable to climate change impacts is one of the main aspects of climate vulnerability assessments. This estimation involves processing, integrating, and analyzing many information sources to provide accurate and timely responses. However, designing this evaluation, examine the information gathered, and reaching agreements among all stakeholders and experts, often requires considerable effort in time, money, and people. In this study, we propose a data fusion strategy to support climate vulnerability assessments by identifying the adaptability of crops in a territory in the short term. This strategy follows the Joint Directors of Laboratories’ data fusion model guidelines. It was evaluated and validated through a case study in Colombia’s upper Cauca river basin. For this purpose, we identified Climate, Soil, Water Quality, Productive Alliances, and Production as the most relevant data sources to be integrated, and using metrics such as Mean IR, SCUMBLE, TCS, among others, we evaluated the combined datasets according to their theoretical complexity. The adaptability of crops in a territory was addressed as a multi-label learning problem, assessing the performance of different multi-label classification and multi-view multi-label classification models with both test and actual data. Comparing the predicted crops with the actual ones, we obtained a 98% similarity without considering crop ranking using the Binary Relevance approach and the Random Forest and XGBoost algorithms. Using a more exhaustive test involving order, we obtained a maximum similarity of 67% applying Binary Relevance and Random Forest.

Multi-Label Learning With Visual-Semantic Embedded Knowledge Graph for Diagnosis of Radiology Imaging

A significant task of automatic diagnosis for radiology imaging, especially for chest X-rays, is to identify disease types, which can be viewed as a multi-label learning problem. Prior state-of-the-art approaches adopted the graph convolutional network to model the correlations among disease labels. However, the utilization of medical reports paired with radiology images is neglected in such approaches. Hence, at least two novel improvements are proposed in this paper. First, disease label embeddings are pre-trained on the total radiology reports, and these semantic features along with encoded X-ray features are fused in a transformer encoder for graph initialization. Second, to expand the representation ability of the graph, extra medical terms from radiology reports are mined and added to the graph model as auxiliary nodes without changing the size of the output space. Experiments conducted on two public chest-X-ray datasets demonstrate the outstanding performance over compared models and the advantages of the proposed improvements.

Optimality Assessments of Classifiers on Single and Multi-labelled Obstetrics Outcome Classification Problems

It is indisputable that clinicians cannot exactly state the outcome of pregnancies through conventional knowledge and methods even as the surge in human knowledge continues. Hence, several computational techniques have been adapted for precise pregnancy outcome (PO) prediction. Obstetric datasets for PO determination exist as single label learning (SLL), multi-label learning (MLL) and multi-target (MTP) problems. There is however no single classifier recommended to optimally satisfy the needs of all the classification types. This work therefore identifies six widely used PO classifiers and investigates their performances in all three classification categories; to find the best performing classifier. Obstetric dataset exposed to input rank analysis via Principal component Analysis, produced thirteen (13) significant features for the experiment. Accuracy, F1-measure and build/test time were used as evaluation metrics. Decision tree (DT) had an average accuracy and F1 score of 89.23% and 88.23% respectively, with 1.0 average rank. Under MLL configuration, average accuracy (91.71%) and F1 score (94.28%) were highest in the random forest (RF) which had a 1.0 average test time rank. Using MTP, DT had an average accuracy of 88.80% and average F1 score of 71.13%, the multi-layered perceptron (MLP) had the best time cost with an average rank value of 2.0. From the results, RF is most optimal in terms of accuracy and average rank value, while DT is the most efficient in terms of time cost. The comparative analysis of global averages of the six base classifiers shows that RF is the most optimal algorithm with an average accuracy of 87.3% given all three data setups in the study. MLP on the other hand had an unexpectedly high time cost, making it unsuitable for similar data classifications if time is the main criterion. It is recommended that the choice of the classifier should either be RF or DT depending on the application domain and whether or not time cost is a major consideration.

Degradation State Partition and Compound Fault Diagnosis of Rolling Bearing Based on Personalized Multilabel Learning

The prognostic and health management (PHM) of rolling bearings has been a popular research area. Since bearing fault is inevitable during degradation, how to improve the PHM performance based on both degradation states and fault types is still an open problem. In this study, two multilabel learning algorithms are proposed for PHM of rolling bearings, named personalized binary relevance (PBR) and hierarchical multilabel K-nearest neighbor (HML-KNN), respectively. Degradation states and fault types are used as the labels of the bearing data so that each sample has a corresponding label sequence, that is to say, the PHM problem is converted to a multilabel learning problem. Both algorithms have a personalized search process, which can not only help samples build a personalized model to improve classification accuracy but also solve the problem of data imbalance between labels. At the same time, the two algorithms also have their own characteristics and focus on different application situations. The PBR algorithm has faster modeling speed, more flexible use, and replaceable subclassifiers. HML-KNN is a high-order algorithm with global information analysis capabilities through the hierarchical processing of data and the conversion of label information. Both methods have achieved good enough results in the XJTU-SY bearing dataset. In order to illustrate the practicality of the algorithm, the experimental part further increases the difficulty. Using only a single faulty sample as the training set to determine the type of compound fault, the algorithms also show high performance. In the era of industrial big data, the depth of data mining and the design of algorithm models will help us better manage equipment health.

Smart Handheld Based Human Activity Recognition Using Multiple Instance Multiple Label Learning

Human activity recognition (HAR) and monitoring is beneficial for many medical applications, such as eldercare and post-trauma rehabilitation after surgery. HAR models based on smartphone’s accelerometer data could provide a convenient and ubiquitous solution to this problem. However, such models are mostly concerned with identifying basic activities such as ‘stand’/‘walk’ and thus the high-level context such as ‘walk in a queue’ for which a set of specific activities is performed remain unnoticed. Consequently, in this paper, we design a HAR framework that can identify a group of activities (rather than a single basic activity) being performed in a time window, thus, enables us to extract more meaningful information about the subject’s overall context. An algorithm is designed to formulate HAR as a multi-instance multi-label (MIML) learning problem. The procedure of generating feature bags of consecutive activity traces having multiple labels is formulated. In this work, the temporal relationship among activities is exploited to obtain a more comprehensive HAR model. Interestingly, the framework is found to completely/partially identify activity sequences that may not even be present in the training dataset. The framework is implemented and found to be working adequately when tested with real dataset collected from 8 users for 12 different activity combinations. MIML-kNN is found to provide maximum average precision (around 90%) even for an unseen test data-set.

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.

Learning From Weakly Labeled Data Based on Manifold Regularized Sparse Model

In multilabel learning, each training example is represented by a single instance, which is relevant to multiple class labels simultaneously. Generally, all relevant labels are considered to be available for labeled data. However, instances with a full label set are difficult to obtain in real-world applications, thus leading to the weakly multilabel learning problem, that is, relevant labels of training data are partially known and many relevant labels are missing, and even abundant training data are associated with an empty label set. To address the problem, we propose a new multilabel method to learn from weakly labeled data. To be specific, an optimization framework is constructed based on the manifold regularized sparse model, in which the correlations among labels and feature structure are considered to model global and local label correlations, thereby achieving discriminative feature analysis for mapping training data to ground-truth label space. Moreover, the proposed method has an excellent mechanism to conduct semisupervised multilabel learning by exploiting training data with the predicted label set of the unlabeled. Experiments on various real-world tasks reveal that the proposed method outperforms some state-of-the-art methods.

Saliency based multiple object cosegmentation by ensemble MIML learning

As an interesting and emerging topic, multiple foreground cosegmentation (MFC) aims at extracting a finite number of common objects from an image collection, which is useful to variety of visual media applications. Although a number of approaches have been proposed to address this problem, many of them are designed with the misleading consistent information, suboptimal image representation, or inefficient segmentation assist and thus still suffer from certain limitations, which reduces their capability in the real-world scenarios. To alleviate these limitations, we propose a novel unsupervised MFC framework, which is composed of three components: unsupervised label generation, saliency based pseudo-annotation and cosegmentation by MIML learning. Specifically, we combine the high-level and low-level feature to represent the proposal objects, and adopt a novel SPAP clustering scheme to obtain more accurate consistent information of common objects. Then the saliency based pseudo-annotation help us reformulate the MFC problem as a Multi-Instance Multi-Label (MIML) learning problem by label propagation. Finally, by introducing a novel ensemble MIML learning scheme, the consistent information of common objects can more efficiently assist the segmentation of the images and get the more accurate segmentation results. We evaluate our framework on widely used public databases including the ICoseg dataset, MSRC dataset and FlickrMFC dataset for single and multiple common object cosegmentation respectively. Comparison results show that the proposed methods reach advanced and efficient performance.

Use of Chou's 5-steps rule to predict the subcellular localization of gram-negative and gram-positive bacterial proteins by multi-label learning based on gene ontology annotation and profile alignment.

To date, many proteins generated by large-scale genome sequencing projects are still uncharacterized and subject to intensive investigations by both experimental and computational means. Knowledge of protein subcellular localization (SCL) is of key importance for protein function elucidation. However, it remains a challenging task, especially for multiple sites proteins known to shuttle between cell compartments to perform their proper biological functions and proteins which do not have significant homology to proteins of known subcellular locations. Due to their low-cost and reasonable accuracy, machine learning-based methods have gained much attention in this context with the availability of a plethora of biological databases and annotated proteins for analysis and benchmarking. Various predictive models have been proposed to tackle the SCL problem, using different protein sequence features pertaining to the subcellular localization, however, the overwhelming majority of them focuses on single localization and cover very limited cellular locations. The prediction was basically established on sorting signals, amino acids compositions, and homology. To improve the prediction quality, focus is actually on knowledge information extracted from annotation databases, such as protein–protein interactions and Gene Ontology (GO) functional domains annotation which has been recently a widely adopted and essential information for learning systems. To deal with such problem, in the present study, we considered SCL prediction task as a multi-label learning problem and tried to label both single site and multiple sites unannotated bacterial protein sequences by mining proteins homology relationships using both GO terms of protein homologs and PSI-BLAST profiles. The experiments using 5-fold cross-validation tests on the benchmark datasets showed a significant improvement on the results obtained by the proposed consensus multi-label prediction model which discriminates six compartments for Gram-negative and five compartments for Gram-positive bacterial proteins.

Emotion Detection in Online Social Networks: A Multilabel Learning Approach

Emotion detection in online social networks (OSNs) can benefit kinds of applications, such as personalized advertisement services, recommendation systems, etc. Conventionally, emotion analysis mainly focuses on the sentence level polarity prediction or single emotion label classification, however, ignoring the fact that emotions might coexist from users’ perspective. To this end, in this work, we address the multiple emotions detection in OSNs from user-level view, and formulate this problem as a multilabel learning problem. First, we discover emotion labels correlations , social correlations , and temporal correlations from an annotated Twitter data set. Second, based on the above observations, we adopt a factor graph-based emotion recognition model to incorporate emotion labels correlations , social correlations , and temporal correlations into a general framework, and detect the multiple emotions based on the multilabel learning approach. Performance evaluation demonstrates that the factor graph-based emotion detection model can outperform the existing baselines.

Multi-View Partial Multi-Label Learning with Graph-Based Disambiguation

In multi-view multi-label learning (MVML), each training example is represented by different feature vectors and associated with multiple labels simultaneously. Nonetheless, the labeling quality of training examples is tend to be affected by annotation noises. In this paper, the problem of multi-view partial multi-label learning (MVPML) is studied, where the set of associated labels are assumed to be candidate ones and only partially valid. To solve the MVPML problem, a two-stage graph-based disambiguation approach is proposed. Firstly, the ground-truth labels of each training example are estimated by disambiguating the candidate labels with fused similarity graph. After that, the predictive model for each label is learned from embedding features generated from disambiguation-guided clustering analysis. Extensive experimental studies clearly validate the effectiveness of the proposed approach in solving the MVPML problem.