Semi-supervised Learning Tasks Research Articles

Multi-head co-training: An uncertainty-aware and robust semi-supervised learning framework

Co-training, an advanced form of self-training, allows multiple base models to learn collaboratively, leading to superior performance in semi-supervised learning tasks. However, its widespread adoption is hindered by high computational costs and intricate design choices. To address these challenges, we present Multi-Head Co-Training, a streamlined and efficient framework that consolidates individual models into a multi-head structure, adding minimal extra parameters. Each classification head in this unified model collaborates with others via a ”Weak and Strong Augmentation” strategy, with diversity organically introduced through robust data augmentation. Consequently, our approach implicitly promotes diversity while incurring only a minor increase in computational overhead, making co-training more accessible. We validate the effectiveness of Multi-Head Co-Training through an empirical study on standard semi-supervised learning benchmarks. For example, our method achieves up to a 3.1% accuracy improvement on the semi-supervised CIFAR dataset compared to recent methods.Recognizing the necessity for more practical performance metrics beyond accuracy, we assess our framework from three additional perspectives: robust generalization, uncertainty, and computational efficiency. To evaluate robust generalization, we expand the conventional SSL experimental setting to a more comprehensive open-set semi-supervised learning scenario. For uncertainty assessment, we conduct experiments on model calibration and selective classification benchmarks. For example, our method achieves up to a 4.3% accuracy improvement on the open-set semi-supervised CIFAR dataset. Our extensive experiments confirm that our proposed framework better captures prediction confidence and uncertainty, rendering it more suitable for SSL deployment in open environments. The code is available at https://github.com/chenmc1996/Multi-Head-Co-Training.

Federated Momentum Contrastive Clustering

Self-supervised representation learning and deep clustering are mutually beneficial to learn high-quality representations and cluster data simultaneously in centralized settings. However, it is not always feasible to gather large amounts of data at a central entity, considering data privacy requirements and computational resources. Federated Learning (FL) has been developed successfully to aggregate a global model while training on distributed local data, respecting the data privacy of edge devices. However, most FL research effort focuses on supervised learning algorithms. A fully unsupervised federated clustering scheme has not been considered in the existing literature. We present federated momentum contrastive clustering (FedMCC), a generic federated clustering framework that can not only cluster data automatically but also extract discriminative representations training from distributed local data over multiple users. In FedMCC, we demonstrate a two-stage federated learning paradigm where the first stage aims to learn differentiable instance embeddings and the second stage accounts for clustering data automatically. The experimental results show that FedMCC not only achieves superior clustering performance but also outperforms several existing federated self-supervised methods for linear evaluation and semi-supervised learning tasks. Additionally, FedMCC can easily be adapted to ordinary centralized clustering through what we call momentum contrastive clustering (MCC). We show that MCC achieves state-of-the-art clustering accuracy results in certain datasets such as STL-10 and ImageNet-10. We also present a method to reduce the memory footprint of our clustering schemes.

Semi-supervised PolSAR Image Change Detection using Similarity Matching

Abstract. The lack of precisely labeled data limits the development of supervised polarimetric synthetic aperture radar (PolSAR) image change detection. Therefore, semi-supervised deep learning methods have recently demonstrated their significant capability for PolSAR image change detection. Similarity Matching (SimMatch) improves the performance of semi-supervised learning tasks across different benchmark datasets and different settings. Introducing SimMatch into the field of PolSAR image change detection can improve the performance of semi-supervised PolSAR image change detection under limited labeled data conditions. Usually, semi-supervision solves the problem of insufficient labeled data by generating pseudo-labels. However, when the pseudo-label method is simply applied, the model will fit on the confident but wrong pseudo-labels, resulting in poor performance. SimMatch offers a solution by requiring the strongly augmented view to share the same semantic similarity (i.e. label prediction) and instance characteristics (i.e. similarity between instances) with a weak augmented view for more intrinsic feature matching. Besides, by using a labeled memory buffer, the two similarities can be isomorphically transformed with each other by introducing the aggregating and unfolding techniques. Therefore, the semantic and instance pseudo-labels can be mutually propagated, and then, the detection performance of the PolSAR image change detection is improved. Experimental results on real PolSAR datasets demonstrated that SimMatch is an effective semi-supervised PolSAR change detection method and its performance surpasses some well-known change detection methods. Compared to the fully-supervised algorithm CWNN, the semi-supervised SimMatch algorithm can improve accuracy by up to 14.4%.

Leveraging Local Variance for Pseudo-Label Selection in Semi-supervised Learning

Semi-supervised learning algorithms that use pseudo-labeling have become increasingly popular for improving model performance by utilizing both labeled and unlabeled data. In this paper, we offer a fresh perspective on the selection of pseudo-labels, inspired by theoretical insights. We suggest that pseudo-labels with a high degree of local variance are more prone to inaccuracies. Based on this premise, we introduce the Local Variance Match (LVM) method, which aims to optimize the selection of pseudo-labels in semi-supervised learning (SSL) tasks. Our methodology is validated through a series of experiments on widely-used image classification datasets, such as CIFAR-10, CIFAR-100, and SVHN, spanning various labeled data quantity scenarios. The empirical findings show that the LVM method substantially outpaces current SSL techniques, achieving state-of-the-art results in many of these scenarios. For instance, we observed an error rate of 5.41% on CIFAR-10 with a single label for each class, 35.87% on CIFAR-100 when using four labels per class, and 1.94% on SVHN with four labels for each class. Notably, the standout error rate of 5.41% is less than 1% shy of the performance in a fully-supervised learning environment. In experiments on ImageNet with 100k labeled data, the LVM also reached state-of-the-art outcomes. Additionally, the efficacy of the LVM method is further validated by its stellar performance in speech recognition experiments.

Stability and Generalization of Hypergraph Collaborative Networks

Graph neural networks have been shown to be very effective in utilizing pairwise relationships across samples. Recently, there have been several successful proposals to generalize graph neural networks to hypergraph neural networks to exploit more complex relationships. In particular, the hypergraph collaborative networks yield superior results compared to other hypergraph neural networks for various semi-supervised learning tasks. The collaborative network can provide high quality vertex embeddings and hyperedge embeddings together by formulating them as a joint optimization problem and by using their consistency in reconstructing the given hypergraph. In this paper, we aim to establish the algorithmic stability of the core layer of the collaborative network and provide generalization guarantees. The analysis sheds light on the design of hypergraph filters in collaborative networks, for instance, how the data and hypergraph filters should be scaled to achieve uniform stability of the learning process. Some experimental results on real-world datasets are presented to illustrate the theory.

Learning to Predict Gradients for Semi-Supervised Continual Learning.

A key challenge for machine intelligence is to learn new visual concepts without forgetting the previously acquired knowledge. Continual learning (CL) is aimed toward addressing this challenge. However, there still exists a gap between CL and human learning. In particular, humans are able to continually learn from the samples associated with known or unknown labels in their daily lives, whereas existing CL and semi-supervised CL (SSCL) methods assume that the training samples are associated with known labels. Specifically, we are interested in two questions: 1) how to utilize unrelated unlabeled data for the SSCL task and 2) how unlabeled data affect learning and catastrophic forgetting in the CL task. To explore these issues, we formulate a new SSCL method, which can be generically applied to existing CL models. Furthermore, we propose a novel gradient learner to learn from labeled data to predict gradients on unlabeled data. In this way, the unlabeled data can fit into the supervised CL framework. We extensively evaluate the proposed method on mainstream CL methods, adversarial CL (ACL), and semi-supervised learning (SSL) tasks. The proposed method achieves state-of-the-art performance on classification accuracy and backward transfer (BWT) in the CL setting while achieving the desired performance on classification accuracy in the SSL setting. This implies that the unlabeled images can enhance the generalizability of CL models on the predictive ability of unseen data and significantly alleviate catastrophic forgetting. The code is available at https://github.com/luoyan407/grad_prediction.git.

Can Graph Neural Networks Go Deeper Without Over-Smoothing? Yes, With a Randomized Path Exploration!

Graph Neural Networks (GNNs) have emerged as one of the most powerful approaches for learning on graph-structured data, even though they are mostly restricted to being shallow in nature. This is because node features tend to become indistinguishable when multiple layers are stacked together. This phenomenon is known as over-smoothing. This paper identifies two core properties of the aggregation approaches that may act as primary causes for over-smoothing. These properties are namely recursiveness and aggregation from higher to lower-order neighborhoods. Thus, we attempt to address the over-smoothing issue by proposing a novel aggregation strategy that is orthogonal to the other existing approaches. In essence, the proposed aggregation strategy combines features from lower to higher-order neighborhoods in a non-recursive way by employing a randomized path exploration approach. The efficacy of our aggregation method is verified through an extensive comparative study on the benchmark datasets w.r.t. the state-of-the-art techniques on semi-supervised and fully-supervised learning tasks.

GPENs: Graph Data Learning With Graph Propagation-Embedding Networks.

Compact representation of graph data is a fundamental problem in pattern recognition and machine learning area. Recently, graph neural networks (GNNs) have been widely studied for graph-structured data representation and learning tasks, such as graph semi-supervised learning, clustering, and low-dimensional embedding. In this article, we present graph propagation-embedding networks (GPENs), a new model for graph-structured data representation and learning problem. GPENs are mainly motivated by 1) revisiting of traditional graph propagation techniques for graph node context-aware feature representation and 2) recent studies on deeply graph embedding and neural network architecture. GPENs integrate both feature propagation on graph and low-dimensional embedding simultaneously into a unified network using a novel propagation-embedding architecture. GPENs have two main advantages. First, GPENs can be well-motivated and explained from feature propagation and deeply learning architecture. Second, the equilibrium representation of the propagation-embedding operation in GPENs has both exact and approximate formulations, both of which have simple closed-form solutions. This guarantees the compactivity and efficiency of GPENs. Third, GPENs can be naturally extended to multiple GPENs (M-GPENs) to address the data with multiple graph structures. Experiments on various semi-supervised learning tasks on several benchmark datasets demonstrate the effectiveness and benefits of the proposed GPENs and M-GPENs.

Coarse-Refined Consistency Learning Using Pixel-Level Features for Semi-Supervised Medical Image Segmentation.

Pixel-level annotations are extremely expensive for medical image segmentation tasks as both expertise and time are needed to generate accurate annotations. Semi-supervised learning (SSL) for medical image segmentation has recently attracted growing attention because it can alleviate the exhausting manual annotations for clinicians by leveraging unlabeled data. However, most of the existing SSL methods do not take pixel-level information (e.g., pixel-level features) of labeled data into account, i.e., the labeled data are underutilized. Hence, in this work, an innovative Coarse-Refined Network with pixel-wise Intra-patch ranked loss and patch-wise Inter-patch ranked loss (CRII-Net) is proposed. It provides three advantages: i) it can produce stable targets for unlabeled data, as a simple yet effective coarse-refined consistency constraint is designed; ii) it is very effective for the extreme case where very scarce labeled data are available, as the pixel-level and patch-level features are extracted by our CRII-Net; and iii) it can output fine-grained segmentation results for hard regions (e.g., blurred object boundaries and low-contrast lesions), as the proposed Intra-Patch Ranked Loss (Intra-PRL) focuses on object boundaries and Inter-Patch Ranked loss (Inter-PRL) mitigates the adverse impact of low-contrast lesions. Experimental results on two common SSL tasks for medical image segmentation demonstrate the superiority of our CRII-Net. Specifically, when there are only 4% labeled data, our CRII-Net improves the Dice similarity coefficient (DSC) score by at least 7.49% when compared to five classical or state-of-the-art (SOTA) SSL methods. For hard samples/regions, our CRII-Net also significantly outperforms other compared methods in both quantitative and visualization results.

Tabular and latent space synthetic data generation: a literature review

The generation of synthetic data can be used for anonymization, regularization, oversampling, semi-supervised learning, self-supervised learning, and several other tasks. Such broad potential motivated the development of new algorithms, specialized in data generation for specific data formats and Machine Learning (ML) tasks. However, one of the most common data formats used in industrial applications, tabular data, is generally overlooked; Literature analyses are scarce, state-of-the-art methods are spread across domains or ML tasks and there is little to no distinction among the main types of mechanism underlying synthetic data generation algorithms. In this paper, we analyze tabular and latent space synthetic data generation algorithms. Specifically, we propose a unified taxonomy as an extension and generalization of previous taxonomies, review 70 generation algorithms across six ML problems, distinguish the main generation mechanisms identified into six categories, describe each type of generation mechanism, discuss metrics to evaluate the quality of synthetic data and provide recommendations for future research. We expect this study to assist researchers and practitioners identify relevant gaps in the literature and design better and more informed practices with synthetic data.

Neighborhood-Regularized Self-Training for Learning with Few Labels

Training deep neural networks (DNNs) with limited supervision has been a popular research topic as it can significantly alleviate the annotation burden. Self-training has been successfully applied in semi-supervised learning tasks, but one drawback of self-training is that it is vulnerable to the label noise from incorrect pseudo labels. Inspired by the fact that samples with similar labels tend to share similar representations, we develop a neighborhood-based sample selection approach to tackle the issue of noisy pseudo labels. We further stabilize self-training via aggregating the predictions from different rounds during sample selection. Experiments on eight tasks show that our proposed method outperforms the strongest self-training baseline with 1.83% and 2.51% performance gain for text and graph datasets on average. Our further analysis demonstrates that our proposed data selection strategy reduces the noise of pseudo labels by 36.8% and saves 57.3% of the time when compared with the best baseline. Our code and appendices will be uploaded to: https://github.com/ritaranx/NeST.

Learning graph representation by aggregating subgraphs via mutual information maximization

Information theory has shown a notable performance in the field of computer vision (CV) and natural language processing (NLP), therefore, many works start to learn better node-level and graph-level representations in the information theory perspective. Previous works have shown great performance by maximizing the mutual information between graph and node representation to capture graph information. However, a simple mixture of information in a single node representation leads to a lack of information related to the graph structure, which leads to the information gap between model and theoretical optimal solution. To solve this problem, we propose to replace the node representation with subgraph representation to reduce the information gap between the model and the optimal case. And to capture enough information of original graph, three operators (information aggregators): attribute-conv, layer-conv and subgraph-conv, are designed to gather information from different aspects, respectively. Moreover, to generate more expressive subgraphs, we propose a universal framework to generate subgraphs autoregressively, which provides a comprehensive understanding of the graph structure in a learnable way. We proposed a Head–Tail negative sampling method to provide more negative samples for more efficient and effective contrastive learning. Moreover, all these components can be plugged into any existed Graph Neural Networks. Experimentally, we achieve new state-of-the-art results in several benchmarks under the unsupervised case. We also evaluate our model on semi-supervised learning tasks and make a fair comparison to state-of-the-art semi-supervised methods.

Pseudo-label self-training model for transfer learning algorithm

When aligning joint distributions between domains, the existing transfer learning algorithms usually assign pseudo labels due to the lack of labels in target domain. However, the noise in pseudo labels will affect the performance of transfer learning. Pseudo-label self-training for transfer learning (PST-TL) model is proposed to generate reliable pseudo labels for target domain and have a wide range of applications in existing algorithms. Pseudo labels are predicted by an ensemble classifier using absolute majority vote, and labels predicted successfully are considered to be high confidence. The training of ensemble classifier applies the self-training of joint pseudo labels strategy, adding strongly stable data to training set of the classifier. The semi-supervised and unsupervised transfer learning tasks in experiment show that the existing transfer learning algorithm can significantly improve the transfer performance after embedded by PST-TL model.

Interpretable Graph Convolutional Network for Multi-View Semi-Supervised Learning

As real-world data become increasingly heterogeneous, multi-view semi-supervised learning has garnered widespread attention. Although existing studies have made efforts towards this and achieved decent performance, they are restricted to shallow models and how to mine deeper information from multiple views remains to be investigated. As a recently emerged neural network, Graph Convolutional Network (GCN) exploits graph structure to propagate label signals and has achieved encouraging performance, and it has been widely employed in various fields. Nonetheless, research on solving multi-view learning problems via GCN is limited and lacks interpretability. To address this gap, in this paper we propose a framework termed Interpretable Multi-view Graph Convolutional Network (IMvGCN <xref ref-type="fn" rid="fn1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">¹</xref> <fn id="fn1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><label>¹</label> Code is available at <uri>https://github.com/ZhihaoWu99/IMvGCN</uri>. </fn> ). We first combine the reconstruction error and Laplacian embedding to formulate a multi-view learning problem that explores the original space from feature and topology perspectives. In light of a series of derivations, we establish a potential connection between GCN and multi-view learning, which holds significance for both domains. Furthermore, we propose an orthogonal normalization method to guarantee the mathematical connection, which solves the intractable problem of orthogonal constraints in deep learning. In addition, the proposed framework is applied to the multi-view semi-supervised learning task. Comprehensive experiments demonstrate the superiority of our proposed method over other state-of-the-art methods.

Sparse norm regularized attribute selection for graph neural networks

Graph Neural Networks (GNNs) have been widely used for graph learning tasks. The main aspect of GNN’s layer-wise message passing is conducting attribute/feature propagation on graph. Most existing GNNs generally conduct feature propagation across all feature dimensions. However, in many real applications, attributes usually contain irrelevant and redundant noise. In this case, attribute/feature selection is desired to extract meaningful features and eliminate noisy ones for GNN’s layer-wise propagation. Based on this observation, in this paper, we combine ℓ2,1/ℓ1-norm regularized attribute selection and GNNs together and propose a novel Attribute selection guided GNNs (AsGNNs) for graph data representation.AsGNNs aim to adaptively select some desired meaningful features/attributes that best serve GNNs. Moreover, an effective optimization framework has also been derived to train the proposed AsGNNs. The proposed AsGNNs provide a general framework which can incorporate any GNNs to conduct feature selection for layer-wise propagation. In this paper, we implement AsGNNs on both graph convolutional network (GCN) and graph attention network (GAT) and develop AsGCN and AsGAT for graph learning. Experimental results on several benchmark datasets demonstrate the effectiveness of the proposed AsGNNs (AsGCN, AsGAT) on semi-supervised learning tasks.

Safe semi-supervised learning using a bayesian neural network

Semi-supervised learning attempts to use a large set of unlabeled data to increase the prediction accuracy of machine learning models when the amount of labeled data is limited. However, in realistic cases, unlabeled data may worsen performance because they contain out-of-distribution (OOD) data that differ from the labeled data. To address this issue, safe semi-supervised deep learning has recently been presented. This study suggests a new safe semi-supervised algorithm that uses an uncertainty-aware Bayesian neural network. Our proposed method, safe uncertainty-based consistency training (SafeUC), uses Bayesian uncertainty to minimize the harmful effects caused by unlabeled OOD examples. The proposed method improves the model’s generalization performance by regularizing the network for consistency against uncertain noise. Moreover, to avoid uncertain prediction results, the proposed method includes a practical inference tip based on a well-calibrated uncertainty. The effectiveness of the proposed method is demonstrated in the experimental results on CIFAR-10 and SVHN by showing that it achieved state-of-the-art performance for all semi-supervised learning tasks with OOD data presence rates.

Heri-Graphs: A Dataset Creation Framework for Multi-Modal Machine Learning on Graphs of Heritage Values and Attributes with Social Media

Values (why to conserve) and Attributes (what to conserve) are essential concepts of cultural heritage. Recent studies have been using social media to map values and attributes conveyed by the public to cultural heritage. However, it is rare to connect heterogeneous modalities of images, texts, geo-locations, timestamps, and social network structures to mine the semantic and structural characteristics therein. This study presents a methodological framework for constructing such multi-modal datasets using posts and images on Flickr for graph-based machine learning (ML) tasks concerning heritage values and attributes. After data pre-processing using pre-trained ML models, the multi-modal information of visual contents and textual semantics are modelled as node features and labels, while their social relationships and spatio-temporal contexts are modelled as links in Multi-Graphs. The framework is tested in three cities containing UNESCO World Heritage properties—Amsterdam, Suzhou, and Venice— which yielded datasets with high consistency for semi-supervised learning tasks. The entire process is formally described with mathematical notations, ready to be applied in provisional tasks both as ML problems with technical relevance and as urban/heritage study questions with societal interests. This study could also benefit the understanding and mapping of heritage values and attributes for future research in global cases, aiming at inclusive heritage management practices. Moreover, the proposed framework could be summarized as creating attributed graphs from unstructured social media data sources, ready to be applied in a wide range of use cases.

Semi-supervised classifier guided by discriminator

Some machine learning applications do not allow for data augmentation or are applied to modalities where the augmentation is difficult to define. Our study aimed to develop a new method in semi-supervised learning (SSL) applicable to various modalities of data (images, sound, text), especially when augmentation is hard or impossible to define, i.e., medical images. Assuming that all samples, labeled and unlabeled, come from the same data distribution, we can say that labeled and unlabeled data sets used in the semi-supervised learning tasks are similar. Based on this observation, the data embeddings created by the classifier should also be similar for both sets. In our method, finding these embeddings is achieved based on two models—classifier and an auxiliary discriminator model, inspired by the Generative Adversarial Network (GAN) learning process. The classifier is trained to build embeddings for labeled and unlabeled datasets to cheat discriminator, which recognizes whether the embedding comes from a labeled or unlabeled dataset. The method was named the DGSSC from Discriminator Guided Semi-Supervised Classifier. The experimental research aimed evaluation of the proposed method on the classification task in combination with the teacher-student approach and comparison with other SSL methods. In most experiments, training the networks with the DGSSC method improves accuracy with the teacher-student approach. It does not deteriorate the accuracy of any experiment.

Jointly optimized ensemble deep random vector functional link network for semi-supervised classification

Randomized neural networks have become more and more attractive recently since they use closed-form solutions for parameter training instead of gradient-based approaches. Among them, the random vector functional link network (RVFL) and its deeper version ensemble deep random vector functional link network (edRVFL) show great performance on both classification and regression tasks. However, the previous research on these two models mainly focuses on the supervised learning area. Although there have been efforts to extend the RVFL network to solve semi-supervised learning problems, the potential of the edRVFL network has not been fully investigated. Therefore, we propose a jointly optimized learning strategy for the edRVFL network (JOSedRVFL) for semi-supervised learning tasks in this paper. The JOSedRVFL network uses an iterative procedure to compute the output weights and consequently predicts the class labels of the unlabeled training data during the training process. In addition, we propose another semi-supervised edRVFL network (SS-edRVFL) using manifold regularization in this work. We then do a brief comparison between these two methods to illustrate their similarities and differences. In the experimental part, we conduct the first set of experiments using the UCI datasets to compare the performance of our proposed semi-supervised algorithms against 11 other classifiers to demonstrate the superior performance of the SS-edRVFL and JOSedRVFL networks. JOSedRVFL achieves the highest accuracy on all 4 datasets while SS-edRVFL takes the second place 3 times which is only worse than JOSedRVFL. Moreover, we apply the proposed methods to real-world applications using the electroencephalography-based emotion recognition dataset to compare the performance of RVFL-based methods (RVFL, SS-RVFL, and JOSRVFL) and their edRVFL counterparts (edRVFL, SS-edRVFL, and JOSedRVFL). Results from this test revealed that the edRVFL-based models (edRVFL, SS-edRVFL, and JOSedRVFL) can obtain higher accuracy than the RVFL-based versions (RVFL, SS-RVFL, and JOSRVFL) with the same learning framework on 45 real-world semi-supervised benchmarks. We then perform the Wilcoxon signed-rank test to show that JOSedRVFL is significantly better than 5 other competitors, which supports our claim that JOSedRVFL can be treated as a superior classifier for semi-supervised classification on both benchmark datasets and real-world applications.

Semisupervised Training of Deep Generative Models for High-Dimensional Anomaly Detection.

Abnormal behaviors in industrial systems may be early warnings on critical events that may cause severe damages to facilities and security. Thus, it is important to detect abnormal behaviors accurately and timely. However, the anomaly detection problem is hard to solve in practice, mainly due to the rareness and the expensive cost to get the labels of the anomalies. Deep generative models parameterized by neural networks have achieved state-of-the-art performance in practice for many unsupervised and semisupervised learning tasks. We present a new deep generative model, Latent Enhanced regression/classification Deep Generative Model (LEDGM), for the anomaly detection problem with multidimensional data. Instead of using two-stage decoupled models, we adopt an end-to-end learning paradigm. Instead of conditioning the latent on the class label, LEDGM conditions the label prediction on the learned latent so that the optimization goal is more in favor of better anomaly detection than better reconstruction that the previously proposed deep generative models have been trained for. Experimental results on several synthetic and real-world small- and large-scale datasets demonstrate that LEDGM can achieve improved anomaly detection performance on multidimensional data with very sparse labels. The results also suggest that both labeled anomalies and labeled normal are valuable for semisupervised learning. Generally, our results show that better performance can be achieved with more labeled data. The ablation experiments show that both the original input and the learned latent provide meaningful information for LEDGM to achieve high performance.