Unlabeled Instances Research Articles

Knowledge-Based Representation Learning for Nucleus Instance Classification From Histopathological Images.

The classification of nuclei in H&E-stained histopathological images is a fundamental step in the quantitative analysis of digital pathology. Most existing methods employ multi-class classification on the detected nucleus instances, while the annotation scale greatly limits their performance. Moreover, they often downplay the contextual information surrounding nucleus instances that is critical for classification. To explicitly provide contextual information to the classification model, we design a new structured input consisting of a content-rich image patch and a target instance mask. The image patch provides rich contextual information, while the target instance mask indicates the location of the instance to be classified and emphasizes its shape. Benefiting from our structured input format, we propose Structured Triplet for representation learning, a triplet learning framework on unlabelled nucleus instances with customized positive and negative sampling strategies. We pre-train a feature extraction model based on this framework with a large-scale unlabeled dataset, making it possible to train an effective classification model with limited annotated data. We also add two auxiliary branches, namely the attribute learning branch and the conventional self-supervised learning branch, to further improve its performance. As part of this work, we will release a new dataset of H&E-stained pathology images with nucleus instance masks, containing 20,187 patches of size 1024 ×1024 , where each patch comes from a different whole-slide image. The model pre-trained on this dataset with our framework significantly reduces the burden of extensive labeling. We show a substantial improvement in nucleus classification accuracy compared with the state-of-the-art methods.

Evaluating the Predictive Performance of Positive- Unlabelled Classifiers

Positive-Unlabelled (PU) learning is a growing area of machine learning that aims to learn classifiers from data consisting of labelled positive and unlabelled instances. Whilst much work has been done proposing methods for PU learning, little has been written on the subject of evaluating these methods. Many popular standard classification metrics cannot be precisely calculated due to the absence of fully labelled data, so alternative approaches must be taken. This short commentary paper critically reviews the main PU learning evaluation approaches and the choice of predictive accuracy measures in 51 articles proposing PU classifiers and provides practical recommendations for improvements in this area.

Experiences with Contrastive Predictive Coding in Industrial Time-Series Classification

Multivariate time-series classification problems are found in many industrial settings; for example, fault detection in a manufacturing process by monitoring sensors signals. It is difficult to obtain large labeled datasets in these settings, for reasons such as limitations in the automatic recording, the need for expert root-cause analysis, and the very limited access to human experts. Therefore, methods that perform classification in a label efficient manner are useful for building and deploying machine learning models in the industrial setting. In this work, we apply a self-supervised learning method called Contrastive Predictive Coding (CPC) to classification tasks on three industrial multivariate time-series datasets. First, the CPC neural network (CPC base) is trained with a large number of unlabeled time-series data instances. Then, a standard supervised classifier such as a multi-layer perception (MLP) is trained on available labeled data using the output embeddings from the pre-trained CPC base. On all three classification datasets, we see increased label efficiency (ability to reach a goal accuracy level with less labeled examples). In the low data regime (10's or few 100's of labeled examples), the CPC pre-trained model achieves high accuracy with up to 15x less labels than a model trained only on labeled data. We also conduct experiments to evaluate the usefulness of CPC pre-trained classifiers as base models to start an active learning loop, and find that uncertainty sampling does not perform significantly better than random sampling during the initial queries.

SemiACO: A semi-supervised feature selection based on ant colony optimization

Feature selection is one of the most efficient procedures for reducing the dimensionality of high-dimensional data by choosing a practical subset of features. Since labeled samples are not always available and labeling data may be time-consuming or costly, the importance of semi-supervised learning becomes apparent. Semi-supervised learning deals with data that includes both labeled and unlabelled instances. This article proposes a method based on Ant Colony Optimization (ACO) for the semi-supervised feature selection problem called SemiACO. The SemiACO algorithm finds features by considering the minimum redundancy between features and the maximum relevancy between the features and the class label. The SemiACO uses a nonlinear heuristic function instead of a linear one. The heuristic learning technique for the ACO heuristic function utilize a Temporal Difference (TD) reinforcement learning algorithm. We characterize the feature selection search space as a Markov Decision Process (MDP), where features indicate the states, and selecting the unvisited features by each ant represents a set of actions. We contrast the efficiency of SemiACO based on various experiments on 14 benchmark datasets, comparing eight semi-supervised feature selection methods.

Predicting Survival Outcomes in the Presence of Unlabeled Data

Many clinical studies require the follow-up of patients over time. This is challenging: apart from frequently observed drop-out, there are often also organizational and financial challenges, which can lead to reduced data collection and, in turn, can complicate subsequent analyses. In contrast, there is often plenty of baseline data available of patients with similar characteristics and background information, e.g., from patients that fall outside the study time window. In this article, we investigate whether we can benefit from the inclusion of such unlabeled data instances to predict accurate survival times. In other words, we introduce a third level of supervision in the context of survival analysis, apart from fully observed and censored instances, we also include unlabeled instances. We propose three approaches to deal with this novel setting and provide an empirical comparison over fifteen real-life clinical and gene expression survival datasets. Our results demonstrate that all approaches are able to increase the predictive performance over independent test data. We also show that integrating the partial supervision provided by censored data in a semi-supervised wrapper approach generally provides the best results, often achieving high improvements, compared to not using unlabeled data.

Class-imbalanced positive instances augmentation via three-line hybrid

The class-imbalance problem is one of the researches of machine learning and data mining. To address the class-imbalance problem, the traditional oversampling algorithm only utilizes the information of the positive instances to generate the synthetic instances with similar characteristics to the minority instances, and there is a problem that the information of the majority instances cannot be used. When the minority instances are too few and too concentrated, such methods suffer from the problem of small disjuncts, resulting in overfitting of the training data. To solve this problem, we incorporate the genetic process of three-line hybrid rice, and a new positive instances augmentation algorithm, i.e., Three-line Hybrid Positive Instance Augmentation (THPIA) is proposed. The THPIA uses the genetic process of three-line hybrid rice to mixup the features of majority-class and minority-class to construct unlabeled instances. Then, the positive instances in the pool of the positive instances are randomly selected to hybridize with the randomly selected unlabeled instances, and the enhanced seed instances of the positive instances are obtained. Finally, a distance constraint is used to prevent the augmented positive instances from generating noisy instances in the negative region. The experimental results on 20 open datasets show that THPIA can effectively utilize the information of the majority instances to enhance the minority instances. Comparing with 7 state-of-the-art methods by Friedman test and Holm’s post-hoc test, THPIA is comparable to CDSMOTE and SMOTE-LOF, and outperforms the remaining 5 state-of-the-art algorithms.

SS R-CNN: Self-Supervised Learning Improving Mask R-CNN for Ship Detection in Remote Sensing Images

Due to the cost of acquiring and labeling remote sensing images, only a limited number of images with the target objects are obtained and labeled in some practical applications, which severely limits the generalization capability of typical deep learning networks. Self-supervised learning can learn the inherent feature representations of unlabeled instances and is a promising technique for marine ship detection. In this work, we design a more-way CutPaste self-supervised task to train a feature representation network using clean marine surface images with no ships, based on which a two-stage object detection model using Mask R-CNN is improved to detect marine ships. Experimental results show that with a limited number of labeled remote sensing images, the designed model achieves better detection performance than supervised baseline methods in terms of mAP. Particularly, the detection accuracy for small-sized marine ships is evidently improved.

A fisher score-based multi-instance learning method assisted by mixture of factor analysis

Multi-instance learning (MIL) is a problem with significant importance due to the existence of ambiguous labels in many practical tasks. The training data in MIL problem is represented by labeled bags containing a various number of unlabeled instances. This inexact label information has brought complexities to the classification of instances and bags. Fisher scores have been utilized to extract fixed-length representation vectors from the MIL bags with the assistance of generative models. However, the generative model utilized by the existing methods might disregard the covariance information. In this paper, we propose a multi-instance learning method (miMFA) utilizing Fisher scores derived from the mixture of factor analysis (MFA) model. The MFA generates nonlinear data with a set of local factor analysis models, while each local model approximates the full covariance Gaussian using latent factors. Thus, the MFA could cover the data distribution and generate Fisher scores effectively. The MFA-based Fisher score is then utilized to form the bag representation. Moreover, a reinforcement vector and the complementary information from auxiliary MIL methods are introduced for performance refinement. The proposed miMFA is evaluated on 11 MIL datasets, and achieves satisfactory results compared with the existing methods.

Space-Aware Reconfiguration

We consider the problem of reconfiguring a set of physical objects into a desired target configuration, a typical (sub)task in robotics and automation, arising in product assembly, packaging, stocking store shelves, and more. In this paper we address a variant, which we call space-aware reconfiguration, where the goal is to minimize the physical space needed for the reconfiguration, while obeying constraints on the allowable collision-free motions of the objects. Since for given start and target configurations, reconfiguration may be impossible, we translate the entire target configuration rigidly into a location that admits a valid sequence of moves, where each object moves in turn just once, along a straight line, from its starting to its target location, so that the overall physical space required by the start, all intermediate, and target configurations for all the objects is minimized. We investigate two variants of space-aware reconfiguration for the often examined setting of n unit discs in the plane, depending on whether the discs are distinguishable (labeled) or indistinguishable (unlabeled). For the labeled case, we propose a representation of size $$O(n^4)$$ of the space of all feasible initial rigid translations, and use it to find, in $$O(n^6)$$ time, a shortest valid translation, or one that minimizes the enclosing disc or axis-aligned rectangle of both the start and target configurations. For the significantly harder unlabeled case, we show that for almost every direction, there exists a translation in this direction that makes the problem feasible. We use this to devise heuristic solutions, where we optimize the translation under stricter notions of feasibility. We present an implementation of such a heuristic, which solves unlabeled instances with hundreds of discs in seconds.

Dlsa: Semi-supervised partial label learning via dependence-maximized label set assignment

Semi-supervised partial label learning is an emerging weakly supervised learning paradigm dealing with partially labeled data and unlabeled data simultaneously. The supervision information is weak and even missing for training instances. Most existing researches simply propagate labels to unlabeled data and focus on inducing a weighted classifier model, ignoring the importance of acquiring reliable label confidences for training instances. In this paper, by label set assignment, the valid label supervision information can be preserved and propagated to unlabeled instances to a greater extent. Then, the training instances are projected into a lower-dimensional feature space. Reliable label confidence can be obtained by maximizing the dependence between the feature description and the associated label information. After that, the predictive model can be induced by preserving the intrinsic manifold structure in both feature space and label space. Accordingly, a novel semi-supervised partial label learning approach via dependence-maximized label set assignment (Dlsa) is proposed, where an accurate and robust model can be induced based on reliable label confidences obtained by label set assignment and dependence-maximized dimensionality reduction. Extensive experiments on five real-world data sets validate the effectiveness and superiority of the proposed method.

In-Depth Research and Analysis of Multilabel Learning Algorithm

Multilabel learning (MLL), as a hot topic in the field of machine learning, has attracted wide attention from many scholars due to its ability to express output space polysemy. In recent years, a large number of achievements about MLL have emerged. Among these achievements, there are several typical issues worthy of attention. Firstly, the correlation among labels plays a key role in improving MLL model training process. Many MLL algorithms try to fully and effectively use the correlation among labels to improve the performance. Secondly, existing MLL evaluation metrics, which is different from those in binary classification, often reflects the generalization performance of MLL classifiers in some aspects. How to choose metrics in algorithms to improve their generalization performance and fairness is another issue that should be concerned. Thirdly, in many practical MLL applications, there are many unlabeled instances due to their labeling cost in training datasets. How to use the wealth information contained in the correlation among unlabeled instances may contribute to reducing of the labeling cost in MLL and improving performance. Fourthly, labels assigned to instances may not be equally descriptive in many applications. How to describe the importance of each label in output space to an instance has become one of research points that many scholars have paid attention to in recent years. This paper reviews the MLL-related research results of correlation among labels, evaluation metric, multilabel semisupervised learning, and label distribution learning (LDL) from a theoretical and algorithmic perspective. Finally, the related research work on MLL is summarized and discussed.

Cross-Project Defect Prediction Method based on Feature Distribution Alignment and Neighborhood Instance Selection

<p>In the practice of software project development, the developed project is a brand-new project. Defect prediction for this type of software project requires the use of other similar projects (i.e. source projects) to collect relevant data to build a defect prediction model, and make defect prediction for the project under development (i.e. target project). However, the prediction model built with the relevant data of the source project cannot achieve the ideal prediction performance when predicting the target project. The main reason is that there is a large data distribution difference between the source project and the target project. The data distribution difference is mainly in the distribution of features between projects and differences between instances. In response to the above problems, starting from both features and instances, a cross-project defect prediction method is proposed. This method first aligns the feature distribution based on the data of the existing target project and the source project data. Then, it selects the labeled instance that is similar to the unlabeled instance in the target project, and finally builds a defect prediction model based on the selected source project instances. Cross-project defect prediction experiments were carried out on the Relink datasets and the Promise datasets. Compared with the classic instance-based cross-project defect prediction method, significant improvements have been made in F-measure and AUC; compared with the prediction of within project defect prediction, it has achieved comparable performance.</p> <p> </p>

GuidedMix-Net: Semi-supervised Semantic Segmentation by Using Labeled Images as Reference

Semi-supervised learning is a challenging problem which aims to construct a model by learning from limited labeled examples. Numerous methods for this task focus on utilizing the predictions of unlabeled instances consistency alone to regularize networks. However, treating labeled and unlabeled data separately often leads to the discarding of mass prior knowledge learned from the labeled examples. In this paper, we propose a novel method for semi-supervised semantic segmentation named GuidedMix-Net, by leveraging labeled information to guide the learning of unlabeled instances. Specifically, GuidedMix-Net employs three operations: 1) interpolation of similar labeled-unlabeled image pairs; 2) transfer of mutual information; 3) generalization of pseudo masks. It enables segmentation models can learning the higher-quality pseudo masks of unlabeled data by transfer the knowledge from labeled samples to unlabeled data. Along with supervised learning for labeled data, the prediction of unlabeled data is jointly learned with the generated pseudo masks from the mixed data. Extensive experiments on PASCAL VOC 2012, and Cityscapes demonstrate the effectiveness of our GuidedMix-Net, which achieves competitive segmentation accuracy and significantly improves the mIoU over 7$\%$ compared to previous approaches.

Semi-supervised shadowed sets for three-way classification on partial labeled data

Shadowed set divides a fuzzy set into three regions through fuzzy-rough transformation to denote acceptance, rejection and uncertain decision. Based on the tri-partition property, shadowed sets are utilized to implement the machine learning methods for uncertain data analysis. The extant uncertain machine learning methods with shadowed sets include the unsupervised clustering on only unlabeled data and the supervised classification on only labeled data. However, for the partial labeled data containing both labeled and unlabeled data instances, the studies of uncertain learning methods with shadowed sets are very limited. Aiming at the requirement, in this paper, we propose a novel semi-supervised shadowed set on partial labeled data and thereby construct semi-supervised shadowed neighborhoods to implement the three-way classification of uncertain data. To construct the semi-supervised shadowed set, we reformulate the objective function of shadowed sets, in which the membership loss in fuzzy-rough transformation is weighted by labeled and unlabeled data. We also analyze the influence of labeled data to the shadowed set construction. Experiments validate that the proposed three-way classification method with semi-supervised shadowed sets is effective to utilize partial labeled data to achieve low-risk uncertain data classification.

Semi-supervised partial multi-label classification via consistency learning

Partial multi-label learning refers to the problem that each instance is associated with a candidate label set involving both relevant and noisy labels. Existing solutions mainly focus on label disambiguation, while ignoring the negative effect of the inconsistency between feature information and label information. Specifically, the existence of completely unlabeled instances makes the estimation of label co-occurrence difficult. To tackle these problems, we propose a novel framework for partial multi-label learning in semi-supervised scenarios by solving the inconsistency between features and labels. In the first stage, the label-level correlation matrix on both labeled and unlabeled instances is derived via Hilbert-Schmidt Independence Criterion (HSIC). The correlation matrix can characterize the label correlation of labeled instances and can propagate the label correlation of unlabeled instances. In the second stage, the proposed framework achieves the training of feature mapping, the recovery of ground-truth labels, and the alleviation of noisy labels in a mutually beneficial manner, and develops an alternative optimization procedure to optimize them. In addition, a nonlinear version is extended by using kernel trick. Experimental studies demonstrate that the proposed methods can achieve competitive superiority against existing well-established methods.

Label augmented and weighted majority voting for crowdsourcing

Crowdsourcing provides an efficient way to obtain multiple noisy labels from different crowd workers for each unlabeled instance. Label integration methods are designed to infer the unknown true label of each instance from its multiple noisy label set. We argue that when the label quality is higher than random classification, the more the number of labels, the better the performance of label integration methods. However, in real-world crowdsourcing scenarios, each instance cannot obtain enough labels for saving costs. To solve this problem, this paper proposes a novel label integration method called label augmented and weighted majority voting (LAWMV). At first, LAWMV uses the K-nearest neighbors (KNN) algorithm to find each instance’s K-nearest neighbors (including itself) and merges their multiple noisy label sets to obtain its augmented multiple noisy label set. Then, the labels from different neighbors are weighted by the distances and the label similarities between each instance and its neighbors. Finally, the integrated label of each instance is inferred by weighted majority voting (MV). The experimental results on 34 simulated and two real-world crowdsourced datasets show that LAWMV significantly outperforms all the other state-of-the-art label integration methods.

反事実伝播: 介入効果推定のための半教師付き学習

Individual treatment effect (ITE) represents the expected improvement in the outcome of taking a particular action to a particular target, and plays important roles in decision making in various domains. However, its estimation problem is difficult because intervention studies to collect information regarding the applied treatments (i.e., actions) and their outcomes are often quite expensive in terms of time and monetary costs. In this study, we consider a semi-supervised ITE estimation problem that exploits more easily-available unlabeled instances to improve the performance of ITE estimation using small labeled data. We combine two ideas from causal inference and semi-supervised learning, namely, matching and label propagation, respectively, to propose Counterfactual Propagation; CP which is the first semi-supervised ITE estimation method. Experiments using semi-real datasets demonstrate that the proposed method can successfully mitigate the data scarcity problem in ITE estimation.

An active semi-supervised deep learning model for human activity recognition

Human activity recognition (HAR), which aims at inferring the behavioral patterns of people, is a fundamental research problem in digital health and ambient intelligence. The application of machine learning methods in HAR has been investigated vigorously in recent years. However, there are still a number of challenges confronting the task, where one significant barrier lies in the longstanding shortage of annotations. To address this issue, we establish a new paradigm for HAR, which integrates active learning and semi-supervised learning into one framework. The main idea is to reduce the annotation cost by actively selecting the most informative samples for annotation, as well as leveraging the unlabelled instances in a semi-supervised way. In particular, we propose to utilize the massive unlabelled data via temporal ensembling of convolutional neural networks (CNN), which yields robust consensus predictions by aggregating the outputs of the training networks on different epochs. We conducted extensive experiments on three public benchmark datasets. The proposed method achieves Macro F1 values of 0.76, 0.45 and 0.91 in a low annotation scenario on PAMAP2, USCHAD and UCIHAR datasets respectively, outperforming a multitude of state-of-the-art deep models. The ablation study proves the effectiveness of the two components of the framework, i.e., active learning-based sample selection and semi-supervised model training with temporal ensembling, in alleviating the issue of insufficient labels. Cross-validation and statistical significance experiments further demonstrate the robustness and generalization ability of the proposed method. The source codes are available at https://github.com/HaixiaBi1982/ActSemiCNNAct.

Bag dissimilarity regularized multi-instance learning

Multi-instance learning (MIL) is able to cope with the weakly supervised problems where the training data is represented by labeled bags consisting of multiple unlabeled instances. Due to its practical significance, MIL has recently drawn increasing attention. Introducing bag representations is an attractive way to learn MIL data. However, it is difficult for the existing MIL methods to utilize both implicit and explicit bag representations simultaneously. In this paper, we propose a bag dissimilarity regularized (BDR) framework that incorporates multiple bag representations regardless of explicitness or implicitness. Here, the implicit bag representations are incorporated into a regularization term that contains the intrinsic geometric information provided by the bag dissimilarities. The regularization term can be added to the objective function of supervised classifiers. An effective method for explicit bag embedding is also proposed, which exploits the Fisher score derived from factor analysis. Finally, we propose two specific BDR methods based on support vector machine and broad learning system. The proposed BDR methods are evaluated on 14 datasets, and have achieved competitive results with limited computation consumption. We also discuss the effectiveness and the characteristics of BDR framework.

Joint Domain Adaption and Pseudo-Labeling for Cross-Project Defect Prediction

Cross-project defect prediction (CPDP) is a hot research topic in recent years. The inconsistent data distribution between source and target projects and lack of labels for most of target instances bring a challenge for defect prediction. Researchers have developed several CPDP methods. However, the prediction performance still needs to be improved. In this paper, we propose a novel approach called Joint Domain Adaption and Pseudo-Labeling (JDAPL). The network architecture consists of a feature mapping sub-network to map source and target instances into a common subspace, followed by a classification sub-network and an auxiliary classification sub-network. The classification sub-network makes use of the label information of labeled instances to generate pseudo-labels. The auxiliary classification sub-network learns to reduce the distribution difference and improve the accuracy of pseudo-labels for unlabeled instances through loss maximization. Network training is guided by the adversarial scheme. Extensive experiments are conducted on 10 projects of the AEEEM and NASA datasets, and the results indicate that our approach achieves better performance compared with the baselines.