Spatial Graph Convolutional Networks

Danel, Tomasz; Spurek, Przemysław; Tabor, Jacek; Śmieja, Marek; Struski, Łukasz; Słowik, Agnieszka; Maziarka, Łukasz

doi:10.1007/978-3-030-63823-8_76

Cited by 57 publications

(49 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As GCN does not consider the ordering of node neighbors, it cannot handle the graph data with such features, e.g., some IIoT datasets like smart factory data have geometric interpretation of the graph that shows an order according to their spatial positions. To address this issue, the spatial GCN was developed [59]. It uses the spatial features of nodes to aggregate information from the neighboring nodes.…”

Section: A Graph Convolutional Networkmentioning

confidence: 99%

Graph Neural Networks for Anomaly Detection in Industrial Internet of Things

Dai

Tang

2022

IEEE Internet Things J.

112

View full text Add to dashboard Cite

The Industrial Internet of Things (IIoT) plays an important role in digital transformation of traditional industries towards Industry 4.0. By connecting sensors, instruments and other industry devices to the Internet, IIoT facilitates the data collection, data analysis, and automated control, thereby improving the productivity and efficiency of the business as well as the resulting economic benefits. Due to the complex IIoT infrastructure, anomaly detection becomes an important tool to ensure the success of IIoT. Due to the nature of IIoT, graph-level anomaly detection has been a promising means to detect and predict anomalies in many different domains such as transportation, energy and factory, as well as for dynamically evolving networks. This paper provides a useful investigation on graph neural networks (GNN) for anomaly detection in IIoTenabled smart transportation, smart energy and smart factory. In addition to the GNN-empowered anomaly detection solutions on point, contextual, and collective types of anomalies, useful datasets, challenges and open issues for each type of anomalies in the three identified industry sectors (i.e., smart transportation, smart energy and smart factory) are also provided and discussed, which will be useful for future research in this area. To demonstrate the use of GNN in concrete scenarios, we show three case studies in smart transportation, smart energy, and smart factory, respectively.

show abstract

Section: A Graph Convolutional Networkmentioning

confidence: 99%

Graph Neural Networks for Anomaly Detection in Industrial Internet of Things

Dai

Tang

2022

IEEE Internet Things J.

112

View full text Add to dashboard Cite

show abstract

“…3a,4a This rotational variance also could be used in data augmentation for enhanced performance of deep-learning models, 14 which is particularly beneficial in the chemical communities that face a dearth of data available for deep-learning chemistry. 7 Figure 1 showed that the 3DGCN model, trained with max pooling, exhibited lowest capability in recognizing the ligand orientation to BACE-1. For example, the A value for LR [±180,z] was 0.430, 0.259, 0.013, and 0.034 in the order of max, sum, avg, and set2set.…”

Section: Bulletin Of the Korean Chemical Societymentioning

confidence: 99%

“…It should be noted that this dichotomous prediction might not be achieved by other GNN models that mainly utilize the 2D molecular attributes as inputs 3a,4a . This rotational variance also could be used in data augmentation for enhanced performance of deep‐learning models, 14 which is particularly beneficial in the chemical communities that face a dearth of data available for deep‐learning chemistry 7 …”

Section: Figurementioning

confidence: 99%

“…2 The graph neural network (GNN) models that utilize the input of 2D molecular graphs, composed of atoms as nodes and bonds as edges, have been applied to predict molecular properties (e.g., solubility and solvation free energy), 3 protein-ligand interactions, 4 and others. 5 In addition, several CNN-based GNN models have recently been proposed to explicitly deal with 3D information on a molecule, such as bond directions and angles, exemplified by 3D graph convolutional network (3DGCN), 6 spectral graph convolutional network (SGCN), 7 directional message passing neural network (DimeNet), 8 and distance-geometric graph convolutional network (DG-GCN). 9 The CNN-based GNNs, like general CNNs, basically consist of convolution layers, pooling layers, and readout layers.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Effects of Pooling Operations on Prediction of Ligand Rotation‐Dependent Protein–Ligand Binding in 3D Graph Convolutional Network

Kim

et al. 2021

Bulletin Korean Chem Soc

View full text Add to dashboard Cite

3D graph convolutional network (3DGCN) is a graph neural network (GNN) variant that utilizes the 3D bond information on molecules for chemical tasks. In 3DGCN, the pooling operation extracts the molecular features from the atomic features for molecule embedding and dimensionality reduction. In this work, we investigated the pooling effects on the performance of 3DGCN in the classification of protein-ligand binding events, especially when a ligand is rotated with respect to a target protein in the 3D Cartesian coordinate. 3DGCN showed the general ability for recognizing the ligand rotation, and its prediction accuracy was found to be pooling-dependent. For example, 3DGCN with max pooling did not recognize the rotations of known active ligands for human β-secretase 1 faithfully, compared with the other pooling operations (sum, avg, and set2set). This work would contribute to augmented architecture evolution of 3DGCN for the chemical tasks that require the 3D molecular information including chirality.

show abstract

“…3D graph conformer encoder. We adopted a simple 3D spatial GNN as the conformer encoder following the strategy of Danel et al 48 , which can learn both the molecular graph representation and spatial distances between atoms in the 3D space. The GNN follows the paradigm of message passing neural networks.…”

Section: Data Preparationmentioning

confidence: 99%

Deep Scaffold Hopping with Multi-modal Transformer Neural Networks

Zheng

Lei²,

Ai³

et al. 2020

Preprint

View full text Add to dashboard Cite

<p>Scaffold hopping, aiming to identify molecules with novel scaffolds but share a similar target biological activity toward known hit molecules, has always been a topic of interest in rational drug design. Computer-aided scaffold hopping would be a valuable tool but at present it suffers from limited search space and incomplete expert-defined rules and thus provides results of unsatisfactory quality. To addree the issue, we describe a fully data-driven model that learns to perform target-centric scaffold hopping tasks. Our deep multi-modal model, DeepHop, accepts a hit molecule and an interest target protein sequence as inputs and design bioisosteric molecular structures to the target compound. The model was trained on 50K experimental scaffold hopping pairs curated from the public bioactivity database, which spans 40 kinases commonly investigated by medicinal chemists. Extensive experiments demonstrated that DeepHop could design more than 70% molecules with improved bioactivity, high 3D similarity, while low 2D scaffold similarity to the template molecules. Our method achieves 2.2 times larger efficiency than state-of-the-art deep learning methods and 4.7 times than rule-based methods. Case studies have also shown the advantages and usefulness of DeepHop in practical scaffold hopping scenario. </p>

show abstract

Spatial Graph Convolutional Networks

Cited by 57 publications

References 7 publications

Graph Neural Networks for Anomaly Detection in Industrial Internet of Things

Graph Neural Networks for Anomaly Detection in Industrial Internet of Things

Effects of Pooling Operations on Prediction of Ligand Rotation‐Dependent Protein–Ligand Binding in 3D Graph Convolutional Network

Deep Scaffold Hopping with Multi-modal Transformer Neural Networks

Contact Info

Product

Resources

About