Common Subspace Research Articles

Fast Unsupervised Cross-modal Hashing With Robust Factorization and Dual Projection

Unsupervised hashing has attracted extensive attention in effectively and efficiently tackling large-scale cross-modal retrieval task. Existing methods typically try to mine the latent common subspace across multimodal data without any category annotation. Despite the exciting progress, there are still three challenges that need to be further addressed: 1) efficiently improving the robustness during latent common subspace learning; 2) harmoniously embedding the intra-modal inherence and inter-modal relevance of multimodal data into Hamming space; and 3) effectively reducing the training time complexity and making the model scalable for large-scale datasets. To well address the above challenges, this study proposes a method named Fast Unsupervised Cross-modal Hashing (FUCH). Specifically, FUCH proposes a semantic-aware collective matrix factorization to learn robust representation via exploiting latent category-specific attributes, and introduces Cauchy loss to measure the factorization process. Accordingly, the above process can effectively embed potential discriminative information into common space, while making the model insensitive for outliers. Moreover, FUCH designs a dual projection learning scheme, which not only learns modality-unique hash functions to excavate individual properties, but also learns modality-mutual hash functions to multimodal correlational properties. Experimental results on three benchmark datasets verify the effectiveness of FUCH under various scenarios.

Multimode Process Monitoring Based on Common and Unique Subspace Decomposition

Multimode Process Monitoring Based on Common and Unique Subspace Decomposition

Flexible control of sequence working memory in the macaque frontal cortex

To memorize a sequence, one must serially bind each item to its rank order. How the brain controls a given input to bind its associated order in sequence working memory (SWM) remains unexplored. Here, we investigated the neural representations underlying SWM control using electrophysiological recordings in the frontal cortex of macaque monkeys performing forward and backward SWM tasks. Separate and generalizable low-dimensional subspaces for sensory and memory information were found within the same frontal circuitry, and SWM control was reflected in these neural subspaces' organized dynamics. Each item at each rank was sequentially entered into a common sensory subspace and, depending on forward or backward task requirement, flexibly and timely sent into rank-selective SWM subspaces. Neural activity in these SWM subspaces faithfully predicted the recalled item and order information in single error trials. Thus, compositional neural population codes with well-orchestrated dynamics in frontal cortex support the flexible control of SWM.

Heterogeneous domain adaptation via incremental discriminative knowledge consistency

Heterogeneous domain adaptation is a challenging problem in transfer learning since samples from the source and target domains reside in different feature spaces with different feature dimensions. The key problem is how to minimize some gaps (e.g., data distribution mismatch) presented in two heterogeneous domains and produce highly discriminative representations for the target domain. In this paper, we attempt to address these challenges with the proposed incremental discriminative knowledge consistency (IDKC) method, which integrates cross-domain mapping, distribution matching, discriminative knowledge preservation, and domain-specific geometry structure consistency into a unified learning model. Specifically, we attempt to learn a domain-specific projection to project original samples into a common subspace in which the marginal distribution is well aligned and the discriminative knowledge consistency is preserved by leveraging the labeled samples from both domains. Moreover, domain-specific structure consistency is enforced to preserve the data manifold from the original space to the common feature space in each domain. Meanwhile, we further apply pseudo labeling to unlabeled target samples based on the feature correlation and retain pseudo labels with high feature correlation coefficients for the next iterative learning. Our pseudo-labeling strategy expands the number of labeled target samples in each category and thus enforces class-discriminative knowledge consistency to produce more discriminative feature representations for the target domain. Extensive experiments on several standard benchmarks for object recognition, cross-language text classification, and digit classification tasks verify the effectiveness of our method.

KNN-GNN: A powerful graph neural network enhanced by aggregating K-nearest neighbors in common subspace

It has been proven that graph neural networks (GNNs) are effective for a variety of graph learning-based applications. Typical GNNs iteratively aggregate messages from immediate neighbors under the homophily assumption. However, there are a larger number of heterophilous networks in our daily life, where the ability of these GNNs is limited. Recently, some GNN models have been proposed to handle networks with heterophily via some key designs like aggregating higher-order neighbors and combining immediate representations. But ”noise” information transmitted from different-order neighbors will be injected into the representations of nodes. In this paper, we propose a new GNN model, called KNN-GNN, to effectively perform the node classification task for networks with various homophily levels. The main idea of KNN-GNN is to learn a comprehensive and accurate representation for each node by integrating not only the local information from its neighborhood but also the non-local information held by its similar nodes decentralized in the network. Specifically, the local information of a node is generated from itself and its 1-hop neighbor. Then, we project all nodes into a common subspace, where similar nodes are desired to be close to each other. The non-local information of a node is gathered by aggregating its K-Nearest Neighbors searched in the common subspace. We evaluate the performance of KNN-GNN on both real and synthetic datasets including networks with diverse homophily levels. The results demonstrate that KNN-GNN outstrips the state-of-the-art baselines. Moreover, the ablation experiments show that the core designs in KNN-GNN play a critical role in node representation learning.

Clean and robust multi-level subspace representations learning for deep multi-view subspace clustering

Deep multi-view subspace clustering has achieved increasing attention owing to its encouraging ability to address the nonlinear multi-view data. Despite significant progress, most existing deep multi-view subspace clustering methods still suffer from the following drawbacks. First, they usually focus on utilizing the last-level feature from the last encoder layer of the auto-encoder related to each view, yet often overlook exploiting the earlier-level features from the earlier encoder layers as well as integrating the multi-level features from multiple encoder layers. Second, when learning the subspace representation from the self-expressive layer, they rarely consider the cleanliness and robustness of the learned subspace representation. To overcome these drawbacks, this paper proposes a novel clean and robust multi-level subspace representations learning for deep multi-view subspace clustering (CRMSRL). Specifically, the auto-encoder is adopted to nonlinearly map each view into the multi-level features, and then multiple self-expressive layers are added between the encoder layers and their corresponding decoder layers to provide multiple information flow paths through the auto-encoder and learn the multi-level subspace representations corresponding to the multi-level features. Consequently, the intricate hierarchical information embedded in the multi-level features is passed to the corresponding multi-level subspace representations. Moreover, with the idea of robust principal component analysis (RPCA), the learned multi-level subspace representations can be further purified to achieve the cleaner and more robust ones. In addition, to excavate the structural consistency among different views, the common layer is designed to interact with different view-specific self-expressive layers and achieve the high-quality common subspace representation, which can be applied to the spectral clustering algorithm to obtain the final clustering results. Experimental results show that CRMSRL outperforms twelve state-of-the-art baseline methods on six benchmark datasets. In particular, CRMSRL achieves 4.5% improvements on the MSRCV1 dataset in terms of accuracy, compared with the best baseline method.

Partial label learning with heterogeneous domain adaptation

Partial label learning (PLL) seeks a classification model with partially labeled (PL) instances. Each PL instance is attached with a candidate label set, with only one being ground-truth. A fundamental assumption of the previous PLL works is that there are sufficient PL instances in the learning process. Nevertheless, this assumption may not always be valid. In real-world scenarios, there may be only a few PL instances available for training. To this end, we introduce a new PLL method with heterogeneous domain adaptation (PLL-HDA). PLL-HDA leverages the knowledge from the source domain to induce a PLL classifier of the target domain, which contains only a few PL instances for training. Firstly, PLL-HDA introduces a common subspace to measure the instances from the source and target domains, where the instances from these domains are described by heterogeneous features of distinct dimensions. Secondly, we augment the features of the common subspace by merging the original features from both domains. Thirdly, a large-margin-based PLL learning system is established on the augmented features. It leads to a simplified dual form, which can be directly resolved by off-the-shelf support vector machine solvers. Lastly, a heuristic framework is proposed to resolve the PLL-HDA classification model. In this framework, the tasks of learning the classifier on the augmented features and identifying the ground-truth labels of target domain are conducted alternately. Extensive experiments on both the controlled PLL and real-world PLL datasets illustrate the superiority of PLL-HDA over the existing PLL methods. Our code is available at: https://github.com/Sux86.

Robust multiple subspaces transfer for heterogeneous domain adaptation

Heterogeneous domain adaptation (HDA) aims to execute knowledge transfer from a source domain to a heterogeneous target domain. Previous works typically inject knowledge from the source and target domain into a common subspace. However, this may lead to the ineffectiveness of knowledge transfer due to the existence of heterogeneity. To overcome this drawback, in this paper, we propose a robust multiple subspaces transfer method for heterogeneous domain adaptation. Specifically, knowledge of two domains is projected into a union of multiple subspaces via a self-expressive model, in which joint distribution alignment and dynamic Laplacian regularization on self-repressive coefficients are included in the loss for characterizing transferability. Moreover, we provide a comprehensive analysis of stability, complexity, generalization, and convergence guarantee for the proposed method. Experiments on benchmark vision and Language datasets verify effectiveness of the proposed approach for heterogeneous domain adaptation.

Deep Multimodal Fusion of Data with Heterogeneous Dimensionality via Projective Networks.

The use of multimodal imaging has led to significant improvements in the diagnosis and treatment of many diseases. Similar to clinical practice, some works have demonstrated the benefits of multimodal fusion for automatic segmentation and classification using deep learning-based methods. However, current segmentation methods are limited to fusion of modalities with the same dimensionality (e.g., 3D+3D, 2D+2D), which is not always possible, and the fusion strategies implemented by classification methods are incompatible with localization tasks. In this work, we propose a novel deep learning-based framework for the fusion of multimodal data with heterogeneous dimensionality (e.g., 3D+2D) that is compatible with localization tasks. The proposed framework extracts the features of the different modalities and projects them into the common feature subspace. The projected features are then fused and further processed to obtain the final prediction. The framework was validated on the following tasks: segmentation of geographic atrophy (GA), a late-stage manifestation of age-related macular degeneration, and segmentation of retinal blood vessels (RBV) in multimodal retinal imaging. Our results show that the proposed method outperforms the state-of-the-art monomodal methods on GA and RBV segmentation by up to 3.10% and 4.64% Dice, respectively.

PhylaGAN: data augmentation through conditional GANs and autoencoders for improving disease prediction accuracy using microbiome data.

Research is improving our understanding of how the microbiome interacts with the human body and its impact on human health. Existing machine learning methods have shown great potential in discriminating healthy from diseased microbiome states. However, Machine Learning based prediction using microbiome data has challenges such as, small sample size, imbalance between cases and controls and high cost of collecting large number of samples. To address these challenges, we propose a deep learning framework phylaGAN to augment the existing datasets with generated microbiome data using a combination of conditional generative adversarial network (C-GAN) and autoencoder. Conditional generative adversarial networks train two models against each other to compute larger simulated datasets that are representative of the original dataset. Autoencoder maps the original and the generated samples onto a common subspace to make the prediction more accurate. Extensive evaluation and predictive analysis was conducted on two datasets, T2D study and Cirrhosis study showing an improvement in mean AUC using data augmentation by 11% and 5% respectively. External validation on a cohort classifying between obese and lean subjects, with a smaller sample size provided an improvement in mean AUC close to 32% when augmented through phylaGAN as compared to using the original cohort. Our findings not only indicate that the generative adversarial networks can create samples that mimic the original data across various diversity metrics, but also highlight the potential of enhancing disease prediction through machine learning models trained on synthetic data. https://github.com/divya031090/phylaGAN.

Transfer component analysis-driven domain adaptation approach for estimating the life of tunnel boring machine disc cutters

Estimating the life of disc cutters in tunnel boring machines is crucial for tunnel excavation efficiency. Accordingly, generating accurate predictions of the state of disc cutters can prevent unexpected downtimes while optimizing cutting performance. The literature presents investigations of the application of machine learning (ML) for estimating the disc cutter life. Nevertheless, diverse excavation conditions and data scarcity have stimulated research on transfer learning (TL) frameworks that tailor models for different domains, that is, domain adaptation (DA), by generating a common subspace in which the discrepancies among them are minimized. This study presents an end-to-end framework to estimate the disc cutter life through transfer component analysis, a DA-based method for projecting different domains by considering excavation project datasets in a common space; then, different ML models are trained, including linear, probabilistic, and neural methods. The results revealed that the Gaussian process regression model was superior to other methods, with R2 = 0.99. In conclusion, TL is a valuable tool for estimating disc cutter life across different domains, enabling accurate predictions and leveraging knowledge to mitigate data scarcity challenges. In addition, an explainable AI analysis is conducted to quantify and identify relevant features impacting negatively on the disc cutter life, evidencing great contribution from the cutter rotation speed, quartz content, screw rate and specific energy.

Discriminative sparse subspace learning with manifold regularization

Common subspace learning methods only utilize local or global structure in feature extraction, and cannot obtain the global optimal discriminative projection matrix. For this reason, this paper proposes a discriminative sparse subspace learning method based on the manifold regularization framework (DSSL-MR), which introduces the graph Laplacian matrix that reflects the intrinsic geometric structure of the sample as a penalty term. DSSL-MR simultaneously uses both sub-manifold and multi-manifold information of samples for obtaining optimal projection to enhance the discriminability of different classes in subspace. DSSL-MR uses the sparse property of the L2,1-norm to constrain the projection matrix, which can eliminate redundant features and select features that are significant for classification. It is a linear supervised method, which belongs to the Fisher discriminant analysis framework. Experimental results on multiple real-world datasets show that the algorithm is very effective in classification and has high recognition rates.

Soft Contrastive Cross-Modal Retrieval

Cross-modal retrieval plays a key role in the Natural Language Processing area, which aims to retrieve one modality to another efficiently. Despite the notable achievements of existing cross-modal retrieval methodologies, the complexity of the embedding space increases with more complex models, leading to less interpretable and potentially overfitting representations. Most existing methods realize outstanding results based on datasets without any error or noise, but that is extremely ideal and leads to trained models lacking robustness. To solve these problems, in this paper, we propose a novel approach, Soft Contrastive Cross-Modal Retrieval (SCCMR), which integrates the deep cross-modal model with soft contrastive learning and smooth label cross-entropy learning to boost common subspace embedding and improve the generalizability and robustness of the model. To confirm the performance and effectiveness of SCCMR, we conduct extensive experiments comparing 12 state-of-the-art methods on three multi-modal datasets by using image–text retrieval as a showcase. The experimental results show that our proposed method outperforms the baselines.

An improved transfer learning approach based on geodesic flow kernel for multiphase batch process soft sensor modeling

For multiphase batch process, the characteristics of process data under various batches differ. Consequently, the soft sensor model built for a particular working condition is inapplicable to other working conditions. Besides, each batch can be divided into several phases whose characteristics are probably different. To address the above problems, a soft sensor model based on phase division and transfer learning strategy is proposed. First, transfer learning strategy is adopted to construct a soft sensor model applicable to various working conditions. Specifically, geodesic flow kernel based on linear local tangent space alignment (LLTSA-GFK) algorithm is designed. By projecting process data to the common manifold subspace, the distribution difference of data between various batches is reduced and the performance of the soft sensor model is enhanced. In addition, sequence-based fuzzy clustering and just-in-time learning (JITL) are adopted to solve the multistage characteristic for batch process. The root-mean-square error ( RMSE), coefficient of determination [Formula: see text], and mean absolute error ( MAE) are adopted to compare the conventional soft sensing approach (i.e., partial least-square regression based on JITL, support vector regression, and back propagation neural network) with the proposed approach. The superiority of the proposed model is verified by a fed-batch penicillin fermentation process.

MLFLHMDA: predicting human microbe-disease association based on multi-view latent feature learning.

A growing body of research indicates that microorganisms play a crucial role in human health. Imbalances in microbial communities are closely linked to human diseases, and identifying potential relationships between microbes and diseases can help elucidate the pathogenesis of diseases. However, traditional methods based on biological or clinical experiments are costly, so the use of computational models to predict potential microbe-disease associations is of great importance. In this paper, we present a novel computational model called MLFLHMDA, which is based on a Multi-View Latent Feature Learning approach to predict Human potential Microbe-Disease Associations. Specifically, we compute Gaussian interaction profile kernel similarity between diseases and microbes based on the known microbe-disease associations from the Human Microbe-Disease Association Database and perform a preprocessing step on the resulting microbe-disease association matrix, namely, weighting K nearest known neighbors (WKNKN) to reduce the sparsity of the microbe-disease association matrix. To obtain unobserved associations in the microbe and disease views, we extract different latent features based on the geometrical structure of microbes and diseases, and project multi-modal latent features into a common subspace. Next, we introduce graph regularization to preserve the local manifold structure of Gaussian interaction profile kernel similarity and add -norms to the projection matrix to ensure the interpretability and sparsity of the model. The AUC values for global leave-one-out cross-validation and 5-fold cross validation implemented by MLFLHMDA are 0.9165 and 0.8942+/-0.0041, respectively, which perform better than other existing methods. In addition, case studies of different diseases have demonstrated the superiority of the predictive power of MLFLHMDA. The source code of our model and the data are available on https://github.com/LiangzheZhang/MLFLHMDA_master.

Refining Graph Structure for Incomplete Multi-View Clustering.

As a challenging problem, incomplete multi-view clustering (MVC) has drawn much attention in recent years. Most of the existing methods contain the feature recovering step inevitably to obtain the clustering result of incomplete multi-view datasets. The extra target of recovering the missing feature in the original data space or common subspace is difficult for unsupervised clustering tasks and could accumulate mistakes during the optimization. Moreover, the biased error is not taken into consideration in the previous graph-based methods. The biased error represents the unexpected change of incomplete graph structure, such as the increase in the intra-class relation density and the missing local graph structure of boundary instances. It would mislead those graph-based methods and degrade their final performance. In order to overcome these drawbacks, we propose a new graph-based method named Graph Structure Refining for Incomplete MVC (GSRIMC). GSRIMC avoids recovering feature steps and just fully explores the existing subgraphs of each view to produce superior clustering results. To handle the biased error, the biased error separation is the core step of GSRIMC. In detail, GSRIMC first extracts basic information from the precomputed subgraph of each view and then separates refined graph structure from biased error with the help of tensor nuclear norm. Besides, cross-view graph learning is proposed to capture the missing local graph structure and complete the refined graph structure based on the complementary principle. Extensive experiments show that our method achieves better performance than other state-of-the-art baselines.

Deep cross-modal subspace clustering with Contrastive Neighbour Embedding

Deep cross-modal clustering has been developing rapidly and attracted considerable attention in recent years. It aims to pursue a consistent subspace from different modalities with deep neural networks and achieves remarkable clustering performance. However, most existing methods do not simultaneously consider the inherently diverse information of each modality and the neighbour geometric structure over cross-modal data, which inevitably degrades the cluster structure revealed by the common subspace. In this paper, we propose a novel method named Deep Cross-Modal Subspace Clustering with Contrastive Neighbour Embedding (DCSC-CNE) to address the above challenge. DCSC-CNE maintains the inherent independence of each modality while concurrently uncovering consistent information across diverse modalities. In addition, we introduce a contrastive neighbour graph in the proposed deep cross-modal subspace clustering framework by performing contrastive learning between positive and negative samples, to highlight the underlying neighbour geometry of the original data and learn discriminative latent (subspace) representations. In this way, DCSC-CNE integrates the consistent-inherent learning and the contrastive neighbour embedding into a unified deep learning framework. Experimental results demonstrate that the proposed method can significantly improve the cross-modal subspace clustering performance compared with state-of-the-art methods on six benchmark datasets.

Anchor-based multi-view subspace clustering with hierarchical feature descent

Multi-view clustering has attracted growing attention owing to its capabilities of aggregating information from various sources and its promising horizons in public affairs. Up till now, many advanced approaches have been proposed in recent literature. However, there are several ongoing difficulties to be tackled. One common dilemma occurs while attempting to align the features of different views. Moreover, due to the fact that many existing multi-view clustering algorithms stem from spectral clustering, this results to cubic time complexity w.r.t. the number of dataset. However, we propose Anchor-based Multi-view Subspace Clustering with Hierarchical Feature Descent(MVSC-HFD) to tackle the discrepancy among views through hierarchical feature descent and project to a common subspace( STAGE 1), which reveals dependency of different views. We further reduce the computational complexity to linear time cost through a unified sampling strategy in the common subspace( STAGE 2), followed by anchor-based subspace clustering to learn the bipartite graph collectively( STAGE 3). Extensive experimental results on public benchmark datasets demonstrate that our proposed model consistently outperforms the state-of-the-art techniques. Our code is publicly available at https://github.com/QiyuanOu/MVSC-HFD/tree/main.

Fusion-competition framework of local topology and global texture for head pose estimation

RGB image and point cloud involve texture and geometric structure, which are widely used for head pose estimation. However, images lack of spatial information, and the quality of point cloud is easily affected by sensor noise. In this paper, a novel fusion-competition framework (FCF) is proposed to overcome the limitations of a single modality. The global texture information is extracted from image and the local topology information is extracted from point cloud to project heterogeneous data into a common feature subspace. The projected texture feature weighted by the channel attention mechanism is embedded into each local point cloud region with different topological features for fusion. The scoring mechanism creates competition among the regions involving local-global fused features to predict final pose with the highest score. According to the evaluation results on the public and our constructed datasets, the FCF improves the estimation accuracy and stability by an average of 13.6 % and 12.7 %, which is compared to nine state-of-the-art methods.

Heterogeneous Domain Adaptation With Generalized Similarity and Dissimilarity Regularization.

Heterogeneous domain adaptation (HDA) aims to address the transfer learning problems where the source domain and target domain are represented by heterogeneous features. The existing HDA methods based on matrix factorization have been proven to learn transferable features effectively. However, these methods only preserve the original neighbor structure of samples in each domain and do not use the label information to explore the similarity and separability between samples. This would not eliminate the cross-domain bias of samples and may mix cross-domain samples of different classes in the common subspace, misleading the discriminative feature learning of target samples. To tackle the aforementioned problems, we propose a novel matrix factorization-based HDA method called HDA with generalized similarity and dissimilarity regularization (HGSDR). Specifically, we propose a similarity regularizer by establishing the cross-domain Laplacian graph with label information to explore the similarity between cross-domain samples from the identical class. And we propose a dissimilarity regularizer based on the inner product strategy to expand the separability of cross-domain labeled samples from different classes. For unlabeled target samples, we keep their neighbor relationship to preserve the similarity and separability between them in the original space. Hence, the generalized similarity and dissimilarity regularization is built by integrating the above regularizers to facilitate cross-domain samples to form discriminative class distributions. HGSDR can more efficiently match the distributions of the two domains both from the global and sample viewpoints, thereby learning discriminative features for target samples. Extensive experiments on the benchmark datasets demonstrate the superiority of the proposed method against several state-of-the-art methods.