Logical Labels Research Articles

Multi-label feature selection based on adaptive label enhancement and class-imbalance-aware fuzzy information entropy

Multi-label feature selection can select representative features to reduce the dimension of data. Since existing multi-label feature selection methods usually suppose that the significance of all labels is consistent, the relationships between samples in the entire label space are generated straightforwardly such that the shape of label distribution and the property of class-imbalance are ignored. To address these issues, we propose a novel multi-label feature selection approach. Based on non-negative matrix factorization (NMF), the similarities between the logical label and label distribution are constrained, which ensures that the shape of label distribution does not deviate from the underlying actual shape to some extent. Further, the relationships between samples in label space and feature space are restricted by graph embedding. Finally, we leverage the properties of label distribution and class-imbalance to generate the relationships between samples in label space and propose a multi-label feature selection approach based on fuzzy information entropy. Eight state-of-the-art methods are compared with the proposed method to validate the effectiveness of our method.

End-to-end multiple object tracking in high-resolution optical sensors of drones with transformer models

Drone aerial imaging has become increasingly important across numerous fields as drone optical sensor technology continues to advance. One critical challenge in this domain is achieving both accurate and efficient multi-object tracking. Traditional deep learning methods often separate object identification from tracking, leading to increased complexity and potential performance degradation. Conventional approaches rely heavily on manual feature engineering and intricate algorithms, which can further limit efficiency. To overcome these limitations, we propose a novel Transformer-based end-to-end multi-object tracking framework. This innovative method leverages self-attention mechanisms to capture complex inter-object relationships, seamlessly integrating object detection and tracking into a unified process. By utilizing end-to-end training, our approach simplifies the tracking pipeline, leading to significant performance improvements. A key innovation in our system is the introduction of a trajectory detection label matching technique. This technique assigns labels based on a comprehensive assessment of object appearance, spatial characteristics, and Gaussian features, ensuring more precise and logical label assignments. Additionally, we incorporate cross-frame self-attention mechanisms to extract long-term object properties, providing robust information for stable and consistent tracking. We further enhance tracking performance through a newly developed self-characteristics module, which extracts semantic features from trajectory information across both current and previous frames. This module ensures that the long-term interaction modules maintain semantic consistency, allowing for more accurate and continuous tracking over time. The refined data and stored trajectories are then used as input for subsequent frame processing, creating a feedback loop that sustains tracking accuracy. Extensive experiments conducted on the VisDrone and UAVDT datasets demonstrate the superior performance of our approach in drone-based multi-object tracking.

Soft-label recover based label-specific features learning

Presently, multi-label classification algorithms are mainly based on positive and negative logical labels, which have achieved good results. However, logical labeling inevitably leads to the label misclassification problem. In addition, missing labels are common in multi-label datasets. Recovering missing labels and constructing soft labels that reflect the mapping relationship between instances and labels is a difficult task. Most existing algorithms can only solve one of these problems. Based on this, this paper proposes a soft-label recover based label-specific features learning (SLR-LSF) to solve the above problems simultaneously. Firstly, the information entropy is used to calculate the confidence matrix between labels, and the membership degree of soft labels is obtained by combining the label density information. Secondly, the membership degree and confidence matrix are combined to construct soft labels, and this process not only solves the problem of missing labels but also obtains soft labels with richer semantic information. Finally, in the process of learning specific label features for soft labels. The local smoothness of the labels learned through stream regularization is complemented by the global label correlation, thus improving the classification performance of the algorithm. To demonstrate the effectiveness of the proposed algorithm, we conduct comprehensive experiments on several datasets.

Feature selection for multi-labeled data based on label enhancement technique and mutual information

In multi-label data, the importance of each label within the logical label vector varies for each sample, and there exist inherent correlations among the labels. However, the logical label vector fails to capture these nuances. Consequently, relying solely on this vector for feature selection in multi-label data results in underutilization of supervisory information. To address this issue, this paper introduces a novel label enhancement algorithm. This algorithm leverages neighborhood information derived from features to transform the logical label vector into a label distribution that effectively reflects label differences and correlations. Subsequently, we propose a feature selection algorithm tailored for multi-label data, which incorporates both the transformed label distribution and mutual information. This algorithm not only accounts for the mutual information between features and label distributions but also captures the mutual information among features themselves. Finally, we evaluate our proposed feature selection algorithm against five state-of-the-art multi-label feature selection algorithms on ten publicly available datasets. The experimental results reveal that our algorithm outperforms its competitors in six distinct evaluation metrics, achieving an average performance improvement of approximately 9%. This substantial enhancement underscores the efficacy of our algorithm in handling complex multi-label data.

Exploiting asymmetric influence between instances for label enhancement

In multi-label learning, label enhancement (LE) aims to recover label distributions from logical labels, thereby reinforcing supervision information in the training set. Existing LE algorithms mainly leverage pairwise similarities between instances to recover label distributions. However, symmetric similarities, which are widely used, are incapable of reflecting individual differences between instances. In this paper, we propose an asymmetric influence between instances to fully explore the information contained in logical labels. First, we build a bipartite network and employ the mass diffusion algorithm on it to calculate the asymmetric influence between instances. Such asymmetry distinguishes the influence of an instance on others from the influence it receives from others. Second, to cope with the new influence, we construct a graph with self-connections. In this way, we allow instances to be influenced by themselves. Extensive experiments on thirteen benchmark datasets demonstrate the superiority of the proposed method over six state-of-the-art LE algorithms.

Label relaxation and shared information for multi-label feature selection

Due to the rapid growth of labels and high-dimensional data, multi-label feature selection has attracted increasing attention. However, two common issues are ignored by existing multi-label feature selection methods: 1) The pseudo-label matrix is constructed by using logical label matrix directly. 2) There is an imbalance of label density in multi-label data. To tackle the mentioned issues, we innovatively propose a new method named Label relaxation and Shared information for Multi-label Feature Selection (LSMFS). Specifically, LSMFS combines a logical label matrix with a non-negative label relaxation matrix to fit a pseudo-label matrix, which is used for learning the correlations of class labels. LSMFS uses the feature weight matrix to capture a shared information of different related labels, which mitigates the influence of low-density labels on feature selection. The above principles are transformed into the objective function of LSMFS, and an alternate iterative optimization algorithm is developed to solve the objective function. Experiments on multi-label datasets from different domains have demonstrated the effectiveness of the proposed method LSMFS. Code is released at https://github.com/HQUF/LSMFS.

Label enhancement via manifold approximation and projection with graph convolutional network

Label enhancement (LE) aims to enrich logical labels into their corresponding label distributions. But existing LE algorithms fail to fully leverage the structural information in the feature space to improve LE learning. To address this key issue, we first apply manifold learning to map the relatedness between low-dimensional feature samples to the label space. Based on the smoothness assumption of manifolds, the implicit correlation between low-dimensional feature and label spaces effectively promotes the LE process, enabling the learning model to accurately capture the mapping relationship between feature and label manifolds. This leads to an LE based on feature representation (LEFR) algorithm. We also propose an LE algorithm based on graph convolutional network (GCN), called LE-GCN. Inspired by the relationship between threshold connections and label connections, we extend GCN to the LE field for the first time to fully exploit the hidden relationships between nodes and labels. By enhancing node information with threshold connections and label connections, the label learning accuracy reaches a new level. Experiments on real-world datasets show that our LEFR and LE-GCN outperform several state-of-the-art LE algorithms.

Weighted cross-modal hashing with label enhancement

Hashing has been extensively studied for large-scale cross-modal retrieval. However, there are still some unexplored issues with existing methods. First, both the initially-provided similarity graph and one-hot label vector only logically describe the yes or no information, which cannot exactly reflect the correlation between instances. Second, all bits in a binary code contribute equally to Hamming distance, which limits its discriminability to preserve refined semantics. To address these limitations, this paper presents a new method, dubbed Weighted Cross-Modal Hashing (WCMH). It introduces the topological structure of categories to enhance the logical labels, i.e., to recover the real-valued distribution of categories. Moreover, we design a weighted Hamming distance that can enlarge the range of distance values by adaptively learning the weights of different hash bits. In light of this, more discriminative information is embedded into hash codes, and the enhanced semantics can be preserved as much as possible. Extensive experiments validate the superior performance of our WCMH, in comparison with some state-of-the-art cross-modal hashing approaches.

Multi-label feature selection based on fuzzy rough sets with metric learning and label enhancement

Multi-label feature selection based on fuzzy rough sets, as a key step of multi-label data preprocessing, has been widely concerned by scholars in recent years. Most of the existing multi-label feature selection algorithms directly treat labels as logical labels and use a single distance metric to describe similarity. However, the variability of label descriptions and the limitations of a single distance measure should not be overlooked. In this paper, we propose a fuzzy rough set model with metric learning and label enhancement. Specifically, we use a kernel membership label enhancement algorithm based on JS divergence to convert logical labels into numerical labels, which not only reflects the importance of different labels, but also takes into account the differences in label distribution. In addition, a multi-metric learning algorithm is proposed for multi-label learning, in which the metric distance function under the label space and feature space can be learned autonomously. Then, based on the proposed model, we propose a novel multi-label feature selection algorithm based on metric learning and fuzzy rough sets. On this basis, a fast multi-label feature selection algorithm is further designed to improve the computational efficiency. In the experiments, compared with other nine algorithms on real-world datasets, the results show the superiority of the proposed algorithm.

Robust multi-label feature learning-based dual space

AbstractMulti-label learning handles instances associated with multiple class labels. The original label space is a logical matrix with entries from the Boolean domain $$\in \left\{ 0,1 \right\} $$ ∈ 0 , 1 . Logical labels cannot show the relative importance of each semantic label to the instances. Most existing methods map the input features to the label space using linear projections considering the label dependencies using a logical label matrix. However, the discriminative features are learned using one-way projection from the feature representation of an instance into a logical label space. There is no manifold in the learning space of logical labels, which limits the potential of learned models. We propose a novel method in multi-label learning to learn the projection matrix from the feature space to the semantic label space and project it back to the original feature space using encoder–decoder deep learning architecture. The key intuition which guides our method is that the discriminative features are identified due to mapping the features back and forth using two linear projections. To the best of our knowledge, this is one of the first attempts to study the ability to reconstruct the original features from the label manifold in multi-label learning. We show that the learned projection matrix identifies a subset of discriminative features across multiple semantic labels. Extensive experiments on real-world datasets show the superiority of the proposed method.

Multi-label Feature selection with adaptive graph learning and label information enhancement

The high dimensionality and complexity of multi-label data make obtaining accurate label sets in practical applications difficult. Noisy data in the labels will affect the model’s classification performance. Existing methods only reconstruct logical labels into semantic labels to mitigate the effect of inherent noise and do not specifically address the noise. In order to solve this problem and eliminate the effect of noise, this paper proposes a multi-label feature selection method with adaptive graph learning and label information enhancement (AGLE). First, each label is represented as a combination of other labels using global label correlation, which can reconstruct the original label matrix to obtain the self-expression label matrix. Based on this, similar labels in two matrices can be considered reliable labels, and not similar may be noise. Therefore, constructing a predictive label matrix to weigh the learning of original and self-expression labels can enhance reliable information. Second, an adaptive graph is utilized to learn global label correlation to ensure that the self-expression label matrix can be continuously optimized. Then, the local label manifold structure is utilized to constrain the adaptive graph for more stable performance. Finally, the adaptive graph and ℓ2,1-norm are combined to guide learning the feature weight matrix. For the proposed method, an iterative algorithm based on alternating optimization is used to compute the weight matrix. Comparing experimental results with several state-of-the-art algorithms demonstrates the effectiveness of the AGLE method for the noise problem.

Predicting Label Distribution From Tie-Allowed Multi-Label Ranking.

Label distribution offers more information about label polysemy than logical label. There are presently two approaches to obtaining label distributions: LDL (label distribution learning) and LE (label enhancement). In LDL, experts must annotate training instances with label distributions, and a predictive function is trained on this training set to obtain label distributions. In LE, experts must annotate instances with logical labels, and label distributions are recovered from them. However, LDL is limited by expensive annotations, and LE has no performance guarantee. Therefore, we investigate how to predict label distribution from TMLR (tie-allowed multi-label ranking) which is a compromise on annotation cost but has good performance guarantees. On the one hand, we theoretically dissect the relationship between TMLR and label distribution. We define EAE (expected approximation error) to quantify the quality of an annotation, provide EAE bounds for TMLR, and derive the optimal range of label distributions corresponding to a given TMLR annotation. On the other hand, we propose a framework for predicting label distribution from TMLR via conditional Dirichlet mixtures. This framework blends the procedures of recovering and learning label distributions end-to-end and allows us to effortlessly encode our knowledge by a semi-adaptive scoring function. Extensive experiments validate our proposal.

Trusted-Data-Guided Label Enhancement on Noisy Labels.

Label distribution covers a certain number of labels, representing the degree to which each label describes the instance. Label enhancement (LE) is a procedure of recovering the label distribution from the logical labels in the training data, the purpose of which is to better depict the label ambiguity through label distribution. However, data annotation inevitably introduces label noise, and it is extremely challenging to implement LE on corrupted labels. To deal with this problem, one way to recover the label distribution from the corrupted labels is to be guided by a small batch of trusted data. In this article, a novel LE method named TALEN is proposed via recovering and progressively refining label distribution guided by trusted data. Specifically, an LE process is applied to the untrusted data to select samples with a clean label. In addition, a combined loss function is designed to train the predictive model for classification. Experiments on datasets with synthetic label noise validate the feasibility of identifying clean labels via the recovered label distribution. Furthermore, experimental results on both synthetic label noise and real-world label noise on image datasets and additional experiments on text datasets show a clear advantage of TALEN over several existing noise-robust learning methods.

Label correlations-based multi-label feature selection with label enhancement

Feature selection, as an important pre-processing technique, can efficiently mitigate the issue of “the curse of dimensionality” by selecting discriminative features especially for multi-label learning, a discriminative feature subset can improve the classification accuracy. The existing feature selection methods for multi-label classification address the problem of label ambiguity by with logical labels. However, the significance of each label is often different in many practical applications. Using logical label to train the model may result in unsatisfactory performance due to not considering the importance of related labels with each sample. To address this issue, a novel multi-label feature selection algorithm is proposed with two-step: label enhancement and label correlations-based feature selection with label enhancement. In the step of label enhancement, a framework of label enhancement based on deep forest is utilized to transform the logical label to label distribution, which contains rich semantic information and then guides a more correct exploration of semantic correlations. In the step of feature selection, a novel multi-label feature selection algorithm is proposed based on label distribution data. Firstly, the samples are divided into multiple different clusters by using spectral clustering in the label space. Then, the label correlations can be reflected by multiple different clusters. Finally, the l2,1-norm is used to construct an objective function to achieve multi-label feature selection. Experimental results demonstrate that competitiveness of the proposed algorithm over six state-of-the-art multi-label feature selection algorithms on eighteen benchmark datasets in terms of six widely accepted evaluation metrics.

Multi-Label Feature Selection via Label Enhancement and Analytic Hierarchy Process

Multi-label feature selection can effectively resolve the challenges of high or even ultra-high dimensionality in multi-label data. However, most existing multi-label feature selection algorithms can only handle a single data type, assume all labels are equally significant and utilize heuristic search strategies, which results in inefficient and relatively unsatisfactory classification accuracy. In view of the above shortcomings, this paper proposes a new multi-label feature selection algorithm that effectively resolves existing algorithms' issues through three innovative procedures. First, a new similarity relation metric is proposed to deal with hybrid data types effectively. Second, a label enhancement algorithm is designed to enhance and transform the logical labels into a label distribution by fully considering the analytic hierarchy process (AHP) embedded with label correlation, which can automatically identify the significance of different labels. Third, a feature weighting evaluation is redesigned in the feature selection process to obtain the optimal feature subset through feature ranking directly. Under these proposed procedures, multi-label feature selection can effectively utilize the abundant semantic information of the label significance and can significantly improve the operating accuracy and efficiency simultaneously. Comparative experiments are conducted on 20 real multi-label datasets with seven state-of-the-art multi-label feature selection algorithms. Experimental results show that the proposed multi-label feature selection algorithm in this paper is about 5–10% better than the algorithms in 80% of the compared datasets.

Graph-Based Class-Imbalance Learning With Label Enhancement.

Class imbalance is a common issue in the community of machine learning and data mining. The class-imbalance distribution can make most classical classification algorithms neglect the significance of the minority class and tend toward the majority class. In this article, we propose a label enhancement method to solve the class-imbalance problem in a graph manner, which estimates the numerical label and trains the inductive model simultaneously. It gives a new perspective on the class-imbalance learning based on the numerical label rather than the original logical label. We also present an iterative optimization algorithm and analyze the computation complexity and its convergence. To demonstrate the superiority of the proposed method, several single-label and multilabel datasets are applied in the experiments. The experimental results show that the proposed method achieves a promising performance and outperforms some state-of-the-art single-label and multilabel class-imbalance learning methods.

Multi-label feature selection based on correlation label enhancement

Feature selection is an effective data preprocessing technique that can effectively alleviate the curse of dimensionality in multi-label learning. The technique selects a subset of features with high discriminative power to maintain or improve the classifier's classification performance. In many practical applications, it is not possible to use multi-label data to express the relative importance of each label versus each sample. Therefore, all relevant or irrelevant labels are measured with the same importance, neglecting the potential relationship between labels and samples. To better account for the relationship between samples and labels, traditional logical labels are enhanced to label distributions. Aiming at the application of latent sample correlation in label enhancement, a new label enhancement algorithm is proposed. First, to extract potential correlation from the sample space, an objective function is proposed and solved. Second, in the process of label enhancement, the correlation is embedded into label distributions. Finally, label distributions are applied to the sparse linear regression model instead of logical labels. Experiments show that proposed method is superior to 5 advanced multi-label feature selection algorithms on 11 multi-label datasets.

A novel granular ball computing-based fuzzy rough set for feature selection in label distribution learning

Label distribution learning is a widely studied supervised learning diagram that can handle the problem of label ambiguity. The increasing size of datasets is accompanied by the disaster of dimensionality, which implies that the arrival of redundant and noisy features undermines the effect of label distribution learning. As a crucial data-preprocessing technique, feature selection is capable of choosing discriminative features. However, due to the complex issue of label ambiguity, traditional feature selection approaches for datasets with logical labels cannot be applied to label distribution data. In this paper, a novel granular ball computing-based fuzzy rough set (GBFRS) is proposed for label distribution feature selection. Specifically, the proposed method is first introduced at the finest granularity, i.e., calculating similarity relations between single data points. Considering that the label ambiguity issue is exacerbated by the label imbalance phenomenon, the relative similarity in label distribution space among samples is computed for better generalization of the model. Then, a robust approximation strategy is devised for the target sample by using its true different and partially different class samples. Finally, with the concept of granular balls, the method explores the similarity relations between balls and samples, and the granular ball computing-based fuzzy rough set method is developed , which is endowed with the ability to simulate the characteristics of large-scale priorities in human thinking and considers local consistency. Extensive experiments conducted on twenty-two datasets show that GBFRS can effectively select more significant features than seven state-of-the-art feature selection algorithms.

Progressive Enhancement of Label Distributions for Partial Multilabel Learning.

Partial multi-label learning (PML) aims to learn a multilabel predictive model from the PML training examples, each of which is associated with a set of candidate labels where only a subset is valid. The common strategy to induce a predictive model is identifying the valid labels in each candidate label set. Nonetheless, this strategy ignores considering the essential label distribution corresponding to each instance as label distributions are not explicitly available in the training dataset. In this article, a novel partial multilabel learning method is proposed to recover the latent label distribution and progressively enhance it for predictive model induction. Specifically, the label distribution is recovered by considering the observation model for logical labels and the sharing topological structure from feature space to label distribution space. Besides, the latent label distribution is progressively enhanced by recovering latent labeling information and supervising predictive model training alternatively to make the label distribution appropriate for the induced predictive model. Experimental results on PML datasets clearly validate the effectiveness of the proposed method for solving partial multilabel learning problems. In addition, further experiments show the high quality of the recovered label distributions and the effectiveness of adopting label distributions for partial multilabel learning.

Multi‐label feature selection based on relative entropy and fuzzy neighborhood mutual discrimination index

AbstractMulti‐label feature selection eliminates irrelevant and redundant features, and then improves the performance of multi‐label classification models. Most multi‐label feature selection algorithms assume that the training set contains logical labels, which means that labels are equally important for instances. However, in practical applications, there are different importances with respect to labels. To solve the problem, a multi‐label feature selection method based on relative entropy and fuzzy neighborhood mutual discriminant index is proposed. Firstly, logical labels are converted to label distribution through label enhancement. Secondly, the neighborhood and relative entropy are introduced into the label distribution, the label neighborhood similarity matrix is constructed to describe the similarity of samples under label space. Finally, the fuzzy neighborhood mutual discrimination index is used to combine the candidate features with the label neighborhood similarity matrix, which is used to judge the distinguishing ability of the candidate features. Comprehensive experiment of eight multi‐label datasets shows that the proposed algorithm has better classification performance than other compared algorithms.