Remaining useful life (RUL) of lithium-ion battery is important to maintain safe and reliable battery operation. Health indicators (HIs) are key features for predicting RUL during battery aging, whereas current method...
详细信息
Remaining useful life (RUL) of lithium-ion battery is important to maintain safe and reliable battery operation. Health indicators (HIs) are key features for predicting RUL during battery aging, whereas current methods only consider their link to capacity. In order to learn the intrinsic connection between the aging features, a RUL prediction method based on multi decoder graph autoencoder (MGAE) and transformer network is proposed, which considers both the link between aging characteristics and the link between aging characteristics and capacity degradation. First, multiple types of aging features are extracted during battery charging and discharging, and HIs are connected into a graph structure by pearson correlation analysis. Thereby, feature information with high correlation is linked through the topology of the graph. Subsequently, the feature graph and feature matrix are input to the graph autoencoder to extract deep features. In graph decoder part, this paper improves to a multi decoder in order to update and select features by the updated graph structure. Finally, new feature matrix is fed into transformer, and RUL prediction is realized by parallel processing through multi-head self-attention. The validity of proposed method is demonstrated by NASA dataset and compared with other advanced methods. The results show that our approach achieves average RE of 0.09 and maintains RMSE of 0.01. Copyright (C) 2024 The Authors. This is an open access article under the CC BY-NC-ND license (https://***/licenses/by-nc-nd/4.0/)
Log-based anomaly detection in software systems is becoming increasingly crucial for monitoring network operations and ensuring system security. Deep learning-based methods are widely used for large-scale log anomaly ...
详细信息
ISBN:
(纸本)9798331530044;9798331530037
Log-based anomaly detection in software systems is becoming increasingly crucial for monitoring network operations and ensuring system security. Deep learning-based methods are widely used for large-scale log anomaly detection due to their capacity to learn complex features. However, current research predominantly treats original logs as simple sequences, ignoring their complex structure and dynamic dependency relationships. Additionally, these methods often rely on extensive labeled data or domain-specific vectors to represent logs for model training, which can be labor-intensive to label manually and ineffective across various domains within a system. To address these challenges, this paper proposes Pre-LogMGAE, a universal masked graph autoencoder (GAE) framework with contrastive learning for self-supervised pre-training for log anomaly detection. In contrast to graph or link reconstruction, Pre-LogMGAE focuses on node feature reconstruction using a masking strategy to reduce the impact of excessive redundant information. Furthermore, we introduce graph Attention Networks (GAT) with the Gated Recurrent Unit (GRU) to incorporate sequence modeling, allowing for capturing long-term and short-term dependencies in log events. We include contrastive learning objectives in fine-tuning to extract diverse features and enhance the algorithm's robustness. Through an extensive evaluation of three real-world datasets and specific case studies with configuration error, Pre-LogMGAE demonstrates superior performance compared to the six baselines, including PCA, IM, DeepLog, LogRobust, LogBERT, and DeepTraLog. This superiority is evident in terms of precision, recall, F1 score, and time efficiency, highlighting Pre-LogMGAE's stability and reliability in anomaly detection. The study aims to improve anomaly detection capabilities in multi-source system logs, offering innovative technical support to enhance system security and reliability.
The widespread application of multi-view graph data has facilitated the development of multi-view graph clustering. Effectively learning multi-view node representations is crucial for discovering inherent patterns in ...
详细信息
ISBN:
(纸本)9798350359329;9798350359312
The widespread application of multi-view graph data has facilitated the development of multi-view graph clustering. Effectively learning multi-view node representations is crucial for discovering inherent patterns in complex systems. However, most existing methods struggle to handle data with both multi-attribute and multi-relation simultaneously, while both attributes and relations are essential for graph clustering. Therefore, this paper proposes a dual-adaptive fusion multiview clustering method based on graph autoencoder. It utilizes multi-view encoders and decoders to encode and reconstruct inputs separately. Additionally, a dual-adaptive fusion module is introduced to integrate multi-view node representations. Through consistency clustering, the proposed method explores the probability distribution consistency among different views, thereby achieving consistent clustering results. Experimental results on three datasets demonstrate the effectiveness of the proposed method in clustering tasks.
Recent applications of pattern recognition techniques to brain connectome-based classification focus on static functional connectivity (FC) neglecting the dynamics of FC over time, and use input connectivity matrices ...
详细信息
ISBN:
(数字)9781665496209
ISBN:
(纸本)9781665496209
Recent applications of pattern recognition techniques to brain connectome-based classification focus on static functional connectivity (FC) neglecting the dynamics of FC over time, and use input connectivity matrices on a regular Euclidean grid. We exploit the graph convolutional networks (GCNs) to learn irregular structural patterns in brain FC networks and propose extensions to capture dynamic changes in network topology. We develop a dynamic graph autoencoder (DyGAE)-based framework to leverage the time-varying topological structures of dynamic brain networks for identification of autism spectrum disorder (ASD). The framework combines a GCN-based DyGAE to encode individual-level dynamic networks into time-varying low-dimensional network embeddings, and classifiers based on weighted fully-connected neural network (FCNN) and long short-term memory (LSTM) to facilitate dynamic graph classification via the learned spatial-temporal information. Evaluation on a large ABIDE resting-state functional magnetic resonance imaging (rs-fMRI) dataset shows that our method outperformed state-of-the-art methods in detecting altered FC in ASD. Dynamic FC analyses with DyGAE learned embeddings also reveal apparent group difference between ASD and healthy controls in network profiles and switching dynamics of brain states.
Drug synergy is a crucial component in drug reuse since it solves the problem of sluggish drug development and the absence of corresponding drugs for several diseases. Predicting drug synergistic relationships can scr...
详细信息
Drug synergy is a crucial component in drug reuse since it solves the problem of sluggish drug development and the absence of corresponding drugs for several diseases. Predicting drug synergistic relationships can screen drug combinations in advance and reduce the waste of laboratory resources. In this research, we proposed a model that utilizes graph autoencoder and convolutional neural networks to predict drug synergy (GAECDS). Our methods include a graph convolutional neural network as an encoder to encode drug features and use a matrix factorization method as a decoder. Multilayer perceptron (MLP) was applied to process cell line features and combine them with drug features. Furthermore, the latent vectors generated during the encoding process are being used to predict drug synergistic scores using a convolutional neural network. By measuring prediction performance using AUC, AUPR, and F1 score, GAECDS superior to other state-of-the-art models. In addition, four pairs of the predicted top 10 drug combinations were found to work well enough for evaluation. The case study shows that the GAECDS approach is useful for identifying potential drug synergy. [graphICS] .
Crowdsourcing deals with combining and aggregating labels from crowds of annotators of unknown reliability. While most works on label aggregation operate under the assumption of independent and identically distributed...
详细信息
ISBN:
(纸本)9798350344820;9798350344813
Crowdsourcing deals with combining and aggregating labels from crowds of annotators of unknown reliability. While most works on label aggregation operate under the assumption of independent and identically distributed data, the present work introduces an algorithm that operates under known data dependencies or correlations. To exploit these dependencies, a novel graph autoencoder-based algorithm is developed that fuses annotator labels for crowdsourced classification tasks. Numerical tests on real data showcase the potential of the proposed approach.
Deep subspace clustering of hyperspectral image (HSI) holds paramount importance for the fine classification of ground elements like land cover and geological units. While graph-based deep subspace learning effectivel...
详细信息
ISBN:
(纸本)9789819755967;9789819755974
Deep subspace clustering of hyperspectral image (HSI) holds paramount importance for the fine classification of ground elements like land cover and geological units. While graph-based deep subspace learning effectively captures feature representations for clustering, it does not fully harness the global spatial information essential for large-scale imagery requiring multiple graph nodes for representation. This study proposes a Superpixel-based Dual-neighborhood Contrastivegraph autoencoder for Deep SubspaceClustering (SDCGSC), which consists of: (1) Superpixel-based dual-neighborhood graph autoencoder. Utilizing superpixels and their dual-neighborhood to construct two graphs facilitates the learning of local spatial information. (2) Contrastive graph learning. The model employs graph feature-level contrastive learning within each graph autoencoder branch and graph-level contrastive learning across branches to cultivate more robust features. (3) Self-expression reconstruction and clustering. Fused features from both branches, incorporating relative global information, are utilized for clustering. To validate the effectiveness of the proposed model, extensive experiments are conducted on multiple benchmark datasets. Results demonstrate that SDCGSC significantly outperformed existing state-of-the-art methods. In conclusion, superpixel and contrastive graph autoencoder-based deep subspace clustering is positive for HSI analysis.
In the last two decades, Light detection and ranging (LiDAR) has been widely employed in forestry applications. Individual tree segmentation is essential to forest management because it is a prerequisite to tree recon...
详细信息
In the last two decades, Light detection and ranging (LiDAR) has been widely employed in forestry applications. Individual tree segmentation is essential to forest management because it is a prerequisite to tree reconstruction and biomass estimation. This paper introduces a general framework to extract individual trees from the LiDAR point cloud based on a graph link prediction problem. First, an undirected graph is generated from the point cloud based on K-nearest neighbors (KNN). Then, this graph is used to train a convolutional autoencoder that extracts the node embeddings to reconstruct the graph. Finally, the individual trees are defined by the separate sets of connected nodes of the reconstructed graph. A key advantage of the proposed method is that no further knowledge about tree or forest structure is required. Seven sample plots from a plantation forest with poplar and dawn redwood species have been employed in the experiments. Though the precision of the experimental results is up to 95 % for poplar species and 92 % for dawn redwood trees, the method still requires more investigations on natural forest types with mixed tree species.
With the increasing automation and integration of equipment, it is urgent to carry out anomaly detection (AD) for the large-scale system to ensure security, in virtue of Industrial Internet of Things (IIoT). Recently ...
详细信息
With the increasing automation and integration of equipment, it is urgent to carry out anomaly detection (AD) for the large-scale system to ensure security, in virtue of Industrial Internet of Things (IIoT). Recently developed intelligent methods focus on component-level diagnosis or detection, resulting in difficulty in the health assessment of system with multisource data coupling. In addition, data-driven methods rarely emphasize the use of knowledge from the real physical system. In this article, we propose a full graph autoencoder to perform one-class group AD for the large-scale IIoT system. The proposed model takes as input data of normal status at training and only comprises several normalized graph convolutional layers, thus it is simple and fast. Different from Euclidean-based methods, the proposed model can handle various irregular structures together. For graph learning, multivariate time series are converted into graph data fused with prior knowledge. To achieve AD, we propose to reconstruct the full graph for the first time to obtain a reliable anomaly score. Besides, we extend a variational model to fully learn the graph representation. Moreover, a graph augmentation operation is employed to improve the accuracy and robustness. The proposed models are evaluated on two multisensor data sets from liquid rocket engine (LRE) systems, and the experimental results demonstrate the effectiveness and generalization of the IIoT system.
In recent years, with the rapid development of the Internet, the amount of data in the network increases exponentially, and people have higher requirements on the quality of network links, which poses new challenges t...
详细信息
ISBN:
(纸本)9789811996962;9789811996979
In recent years, with the rapid development of the Internet, the amount of data in the network increases exponentially, and people have higher requirements on the quality of network links, which poses new challenges to network management and analysis. Network link status classification can be used to predict network link status categories by using the current network topology information and feature information, facilitating network management and analysis. However, most of the existing models consider to predict the time sequence information of the link status without considering the topology information and node attributes of the network. Therefore, we design a network link status classification model based on graph autoencoder for computer network scenarios. The attention mechanism is introduced into the encoder, and then the node vector is spliced into edge vector, which is then input into the multi-layer perceptron for classification. Finally, the feasibility and validity of the link status classification model are verified on two datasets.
暂无评论