Добірка наукової літератури з теми "Variational graph auto-Encoder (VGAE)"

Graph convolutional networks (GCN) have achieved promising performance in attributed graph clustering and semi-supervised node classification because it is capable of modeling complex graphical structure, and jointly learning both features and relations of nodes. Inspired by the success of unsupervised learning in the training of deep models, we wonder whether graph-based unsupervised learning can collaboratively boost the performance of semi-supervised learning. In this paper, we propose a multi-task graph learning model, called collaborative graph convolutional networks (CGCN). CGCN is composed of an attributed graph clustering network and a semi-supervised node classification network. As Gaussian mixture models can effectively discover the inherent complex data distributions, a new end to end attributed graph clustering network is designed by combining variational graph auto-encoder with Gaussian mixture models (GMM-VGAE) rather than the classic k-means. If the pseudo-label of an unlabeled sample assigned by GMM-VGAE is consistent with the prediction of the semi-supervised GCN, it is selected to further boost the performance of semi-supervised learning with the help of the pseudo-labels. Extensive experiments on benchmark graph datasets validate the superiority of our proposed GMM-VGAE compared with the state-of-the-art attributed graph clustering networks. The performance of node classification is greatly improved by our proposed CGCN, which verifies graph-based unsupervised learning can be well exploited to enhance the performance of semi-supervised learning.

Research on vehicle trajectory prediction based on road monitoring video data often utilizes a global map as an input, disregarding the fact that drivers rely on the road structures observable from their own positions for path planning. This oversight reduces the accuracy of prediction. To address this, we propose the CVAE-VGAE model, a novel trajectory prediction approach. Initially, our method transforms global perspective map data into vehicle-centric map data, representing it through a graph structure. Subsequently, Variational Graph Auto-Encoders (VGAEs), an unsupervised learning framework tailored for graph-structured data, are employed to extract road environment features specific to each vehicle’s location from the map data. Finally, a prediction network based on the Conditional Variational Autoencoder (CVAE) structure is designed, which first predicts the driving endpoint and then fits the complete future trajectory. The proposed CVAE-VGAE model integrates a self-attention mechanism into its encoding and decoding modules to infer endpoint intent and incorporate road environment features for precise trajectory prediction. Through a series of ablation experiments, we demonstrate the efficacy of our method in enhancing vehicle trajectory prediction metrics. Furthermore, we compare our model with traditional and frontier approaches, highlighting significant improvements in prediction accuracy.

Variational Graph Autoencoder (VGAE) has recently gained traction for learning representations on graphs. Its inception has allowed models to achieve state-of-the-art performance for challenging tasks such as link prediction, rating prediction, and node clustering. However, a fundamental flaw exists in Variational Autoencoder (VAE)-based approaches. Specifically, merely minimizing the loss of VAE increases the deviation from its primary objective. Focusing on Variational Graph Autoencoder for Community Detection (VGAECD) we found that optimizing the loss using the stochastic gradient descent often leads to sub-optimal community structure especially when initialized poorly. We address this shortcoming by introducing a dual optimization procedure. This procedure aims to guide the optimization process and encourage learning of the primary objective. Additionally, we linearize the encoder to reduce the number of learning parameters. The outcome is a robust algorithm that outperforms its predecessor.

In a high-speed rail system, the driver repeatedly adjusts the train’s speed and traction while driving, causing a high level of energy consumption. This also leads to the instability of the train’s operation, affecting passengers’ experiences and the operational efficiency of the system. To solve this problem, we propose a variational graph auto-encoder (VGAE) model using a neural network to learn the posterior distribution. This model can effectively capture the correlation between the components of a high-speed rail system and simulate drivers’ operating state accurately. The specific traction control is divided into two parts. The first part employs an algorithm based on the K-Nearest Neighbors (KNN) algorithm and undersampling to address the negative impact of imbalanced quantities in the training dataset. The second part utilizes a variational graph autoencoder to derive the initial traction control of drivers, thereby predicting the energy performance of the drivers’ operation. An 83,786 m long high-speed train driving section is used as an example for verification. By using a confusion matrix for our comparative analysis, it was concluded that the energy consumption is approximately 18.78% less than that of manual traction control. This shows the potential and effect of the variational graph autoencoder model for optimizing energy consumption in high-speed rail systems.

Video summarization aims to identify important segments in a video and merge them into a concise representation, enabling users to comprehend the essential information without watching the entire video. Graph structure-based video summarization approaches ignore the issue of redundant adjacency matrix. To address this issue, this paper proposes a video summary generation model based on graph structure reconstruction (VOGNet), in which the model first adopts a variational graph auto-encoders (VGAE) to reconstruct the graph structure to remove redundant information in the graph structure; followed by using the reconstructed graph structure in a graph attention network (GAT), allocating different weights to different shot features in the neighborhood; and lastly, in order to avoid the loss of information during the training of the model, a feature fusion approach is proposed to combine the training obtained shot features with the original shot features as the shot features for generating the summary. We perform extensive experiments on two standard datasets, SumMe and TVSum, and the experimental results demonstrate the effectiveness and robustness of the proposed model.

Ocean wireless sensor networks (OWSNs) play an important role in marine environment monitoring, underwater target tracking, and marine defense. OWSNs not only monitor the surface information in real time but also act as an important relay layer for underwater sensor networks to establish data communication between underwater sensors and ship-based base stations, land-based base stations, and satellites. The destructive resistance of OWSNs is closely related to the marine environment where they are located. Affected by the dynamics of seawater, the location of nodes is extremely easy to shift, resulting in the deterioration of the connectivity of the OWSNs and the instability of the network topology. In this paper, a novel topology optimization model of OWSNs based on the idea of link prediction by cascading variational graph auto-encoders and adaptive multilayer filter (VGAE-AMF) was proposed, which attenuates the extent of damage after the network is attacked, extracts the global features of OWSNs by graph convolutional network (GCN) to obtain the graph embedding vector of the network so as to decode and generate a new topology, and finally, an adaptive multilayer filter (AMF) is used to achieve topology control at the node level. Simulation experiment results show that the robustness index of the optimized network is improved by 39.65% and has good invulnerability to both random and deliberate attacks.

Abstract Background: Understanding the composition of cancer-associated stroma (CAS) is vital, as the number and location of immune cells and fibroblasts, as well as the degree of extracellular matrix deposition, have implications for cancer progression and response to treatment, including in non-small cell lung cancer (NSCLC). Manual analysis of CAS does not fully describe the stromal milieu, especially from a spatial perspective, and is highly subjective. To this end, we have developed an unsupervised machine learning (ML) model to characterize the CAS in NSCLC from hematoxylin and eosin (H&E) stained whole slide images (WSI) at scale. Methods: PathExploreTM models were deployed to predict stromal tissue and cell types, while another ML model was used to detect collagen fibers from H&E stained WSIs from the TCGA LUAD (N=536) and LUSC (N=464) datasets. Stroma was divided into small regions (median = 0.02 mm2), and 88 features characterizing cell distribution, tissue composition and fiber density were extracted from each region. Graphs were generated connecting neighboring regions (nodes), and an unsupervised variational graph auto-encoder (VGAE) model was trained to learn 8 latent features through dimensionality reduction. Stromal phenotypes were then derived from the latent features using k-means clustering. The fraction of each phenotype in the stroma was correlated against immune- and stroma-related gene expression signatures (GES) and overall survival (OS). Results: Deployment of VGAE on LUAD and LUSC WSIs revealed three distinct stromal phenotypes - P0, P1 and P2. Fibroblast density was elevated in P0 and P1 regions (p<0.001), immune cell density was elevated in P2 regions (p<0.001), and collagen fiber intensity was highest in P1 regions (p<0.001). P2 enrichment was correlated with elevated expression of the T cell-inflamed gene expression profile (TGEP; Spearman ρ = 0.43 in LUAD; ρ = 0.27 in LUSC) and with improved OS (HR = 0.696; 95% CIs: 0.571-0.847 in LUSC). Conversely, P1 enrichment was positively associated with a transforming growth factor-β-induced cancer associated fibroblast GES (TGFβ-CAF: ρ = 0.19 in LUAD and ρ = 0.12 in LUSC) and poor OS (HR = 1.358; 95% CIs: 1.149-1.603 in LUSC). These phenotypes are consistent with fibroblast-enriched, collagen-depleted stroma (P0), collagen-rich, fibroblast-enriched tumor-promoting stroma (P1), and immune cell-enriched, tumor-suppressive stroma (P2). Conclusions: We describe an unsupervised, data-driven method of predicting stromal regions with discrete patterns of cell composition and collagen deposition in NSCLC. This approach identified three phenotypes of NSCLC stroma. These results highlight the ability of ML models to characterize and find meaningful patterns within the cell, tissue, and matrix components of a tumor. This work provides further evidence of the potential of ML to discover novel precision medicine biomarkers in NSCLC. Citation Format: Neel Patel, Nhat Le, Tan Nguyen, Fedaa Najdawi, Sandhya Srinivasan, Adam Stanford-Moore, Deeksha Kartik, Jun Zhang, Jacqueline Brosnan-Cashman, Robert Egger, Justin Lee, Matthew Bronnimann. Unsupervised detection of stromal phenotypes with distinct fibrogenic and inflamed properties in NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4912.

The (variational) graph auto-encoder and its variants have been popularly used for representation learning on graph-structured data. While the encoder is often a powerful graph convolutional network, the decoder reconstructs the graph structure by only considering two nodes at a time, thus ignoring possible interactions among edges. On the other hand, structured prediction, which considers the whole graph simultaneously, is computationally expensive. In this paper, we utilize the well-known triadic closure property which is exhibited in many real-world networks. We propose the triad decoder, which considers and predicts the three edges involved in a local triad together. The triad decoder can be readily used in any graph-based auto-encoder. In particular, we incorporate this to the (variational) graph auto-encoder. Experiments on link prediction, node clustering and graph generation show that the use of triads leads to more accurate prediction, clustering and better preservation of the graph characteristics.

As most networks come with some content in each node, attributed network embedding has aroused much research interest. Most existing attributed network embedding methods aim at learning a fixed representation for each node encoding its local proximity. However, those methods usually neglect the global information between nodes distant from each other and distribution of the latent codes. We propose Structural Adversarial Variational Graph Auto-Encoder (SAVGAE), a novel framework which encodes the network structure and node content into low-dimensional embeddings. On one hand, our model captures the local proximity and proximities at any distance of a network by exploiting a high-order proximity indicator named Rooted Pagerank. On the other hand, our method learns the data distribution of each node representation while circumvents the side effect its sampling process causes on learning a robust embedding through adversarial training. On benchmark datasets, we demonstrate that our method performs competitively compared with state-of-the-art models.

Cette thèse est réalisée dans le cadre du projet BPI DeepTech, en collaboration avec la société Fair&Smart, veillant principalement à la protection des données personnelles conformément au Règlement Général sur la Protection des Données (RGPD). Dans ce contexte, nous avons proposé un modèle neuronal profond pour l'extraction d'informations dans les documents administratifs semi-structurés (DSSs). En raison du manque de données d'entraînement publiques, nous avons proposé un générateur artificiel de DSSs qui peut générer plusieurs classes de documents avec une large variation de contenu et de mise en page. Les documents sont générés à l'aide de variables aléatoires permettant de gérer le contenu et la mise en page en respectant des contraintes visant à garantir leur proximité avec des documents réels. Des métriques ont été introduites pour évaluer la diversité des DSSs générés en termes de contenu et de mise en page. Les résultats de l'évaluation ont montré que les jeux de données générés pour trois types de DSSs (fiches de paie, tickets de caisse et factures) présentent un degré élevé de diversité, ce qui permet d'éviter le sur-apprentissage lors de l'entraînement des systèmes d'extraction d'informations. En s'appuyant sur le format spécifique des DSSs, constitué de paires de mots (mots-clés, informations) situés dans des voisinages proches spatialement, le document est modélisé sous forme de graphe où les nœuds représentent les mots et les arcs, les relations de voisinage. Le graphe est incorporé dans un réseau d'attention à graphe (GAT) multi-couches (Multi-GAT). Celui-ci applique le mécanisme d'attention multi-têtes permettant d'apprendre l'importance des voisins de chaque mot pour mieux le classer. Une première version de ce modèle a été utilisée en mode supervisé et a obtenu un score F1 de 96 % sur deux jeux de données de factures et de fiches de paie générées, et de 89 % sur un ensemble de tickets de caisse réels (SROIE). Nous avons ensuite enrichi le Multi-GAT avec un plongement multimodal de l'information au niveau des mots (avec des composantes textuelle, visuelle et positionnelle), et l'avons associé à un auto-encodeur variationnel à graphe (VGAE). Ce modèle fonctionne en mode semi-supervisé, capable d'apprendre à partir des données annotées et non annotées simultanément. Pour optimiser au mieux la classification des nœuds du graphe, nous avons proposé un semi-VGAE dont l'encodeur partage ses premières couches avec le classifieur Multi-GAT. Cette optimisation est encore renforcée par la proposition d'une fonction de perte VGAE gérée par la perte de classification. En utilisant une petite base de données non annotées, nous avons pu améliorer de plus de 3 % le score F1 obtenu sur un ensemble de factures générées. Destiné à fonctionner dans un environnement protégé, nous avons adapté l'architecture du modèle pour son chiffrement homomorphe. Nous avons étudié une méthode de réduction de la dimensionnalité du modèle Multi-GAT. Ensuite, nous avons proposé une approche d'approximation polynomiale des fonctions non-linéaires dans le modèle. Pour réduire la dimension du modèle, nous avons proposé une méthode de fusion de caractéristiques multimodales qui nécessite peu de paramètres supplémentaires et qui réduit les dimensions du modèle tout en améliorant ses performances. Pour l'adaptation au chiffrement, nous avons étudié des approximations polynomiales de degrés faibles aux fonctions non-linéaires avec une utilisation des techniques de distillation de connaissance et de fine tuning pour mieux adapter le modèle aux nouvelles approximations. Nous avons pu minimiser la perte lors de l'approximation d'environ 3 % pour deux jeux de données de factures ainsi qu'un jeu de données de fiches de paie et de 5 % pour SROIE
This thesis is being carried out as part of the BPI DeepTech project, in collaboration with the company Fair&Smart, primarily looking after the protection of personal data in accordance with the General Data Protection Regulation (RGPD). In this context, we have proposed a deep neural model for extracting information in semi-structured administrative documents (SSDs). Due to the lack of public training datasets, we have proposed an artificial generator of SSDs that can generate several classes of documents with a wide variation in content and layout. Documents are generated using random variables to manage content and layout, while respecting constraints aimed at ensuring their similarity to real documents. Metrics were introduced to evaluate the content and layout diversity of the generated SSDs. The results of the evaluation have shown that the generated datasets for three SSD types (payslips, receipts and invoices) present a high diversity level, thus avoiding overfitting when training the information extraction systems. Based on the specific format of SSDs, consisting specifically of word pairs (keywords-information) located in spatially close neighborhoods, the document is modeled as a graph where nodes represent words and edges, neighborhood connections. The graph is fed into a multi-layer graph attention network (Multi-GAT). The latter applies the multi-head attention mechanism to learn the importance of each word's neighbors in order to better classify it. A first version of this model was used in supervised mode and obtained an F1 score of 96% on two generated invoice and payslip datasets, and 89% on a real receipt dataset (SROIE). We then enriched the multi-GAT with multimodal embedding of word-level information (textual, visual and positional), and combined it with a variational graph auto-encoder (VGAE). This model operates in semi-supervised mode, being able to learn on both labeled and unlabeled data simultaneously. To further optimize the graph node classification, we have proposed a semi-VGAE whose encoder shares its first layers with the multi-GAT classifier. This is also reinforced by the proposal of a VGAE loss function managed by the classification loss. Using a small unlabeled dataset, we were able to improve the F1 score obtained on a generated invoice dataset by over 3%. Intended to operate in a protected environment, we have adapted the architecture of the model to suit its homomorphic encryption. We studied a method of dimensionality reduction of the Multi-GAT model. We then proposed a polynomial approximation approach for the non-linear functions in the model. To reduce the dimensionality of the model, we proposed a multimodal feature fusion method that requires few additional parameters and reduces the dimensions of the model while improving its performance. For the encryption adaptation, we studied low-degree polynomial approximations of nonlinear functions, using knowledge distillation and fine-tuning techniques to better adapt the model to the new approximations. We were able to minimize the approximation loss by around 3% on two invoice datasets as well as one payslip dataset and by 5% on SROIE

Most recent graph clustering methods rely on pretraining graph auto-encoders using self-supervision techniques (pretext task) and finetuning based on pseudo-supervision (main task). However, the transition from self-supervision to pseudo-supervision has never been studied from a geometric perspective. Herein, we establish the first systematic exploration of the latent manifolds' geometry under the deep clustering paradigm; we study the evolution of their intrinsic dimension and linear intrinsic dimension. We find that the embedded manifolds undergo coarse geometric transformations under the transition regime: from curved low-dimensional to flattened higher-dimensional. Moreover, we find that this inappropriate flattening leads to clustering deterioration by twisting the curved structures. To address this problem, which we call Feature Twist, we propose a variational graph auto-encoder that can smooth the local curves before gradually flattening the global structures. Our results show a notable improvement over multiple state-of-the-art approaches by escaping Feature Twist.

Graph neural networks have been widely used for a variety of learning tasks. Link prediction is a relatively under-studied graph learning task, with current state-of-the-art models based on one- or two-layer shallow graph auto-encoder (GAE) architectures. In this paper, we overcome the limitation of current methods for link prediction of non-Euclidean network data, which can only use shallow GAEs and variational GAEs. Our proposed methods innovatively incorporate standard auto-encoders (AEs) into the architectures of GAEs to capitalize on the intimate coupling of node and edge information in complex network data. Empirically, extensive experiments on various datasets demonstrate the competitive performance of our proposed approach. Theoretically, we prove that our deep extensions can inclusively express multiple polynomial filters with different orders. The codes of this paper are available at https://github.com/xinxingwu-uk/DGAE.