EGAT: Edge-Featured Graph Attention Network

Wang, Ziming; Chen, Haopeng

doi:10.1007/978-3-030-86362-3_21

Cited by 38 publications

(21 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Graph Attention Networks (GAT) were first introduced by Veličković et al [29], where attention was used to update nodes in a graph. Edge features might also be important for certain applications, and some studies have therefore tried to extend the concept of GATs to facilitate such features [26]. Incorporating attention into the model, our first aim was to be able to elegantly handle a variable number of upstream neighbours.…”

Section: Gnn Model With Attentionmentioning

confidence: 99%

“…nodes, the input graph was transformed to a temporary edge-graph, where an edge e ij sends to e jk via the node v j . The edge-graph mapping was inspired by the works in [26], producing a new graph with edge features mapped to nodes. Different to [26], the proposed mapping works for directed graphs, as visualised for an arbitrary graph in Fig.…”

Section: Gnn Model With Attentionmentioning

confidence: 99%

“…The edge-graph mapping was inspired by the works in [26], producing a new graph with edge features mapped to nodes. Different to [26], the proposed mapping works for directed graphs, as visualised for an arbitrary graph in Fig. 6.…”

Section: Gnn Model With Attentionmentioning

confidence: 99%

“…With regards to wind, attention mechanisms have mainly been used to extract temporal tendencies in time-series data, and thereby improve RNN-based forecasting models [23,24,25]. Due to the success of attention-based sequence models, there has been a surge in the application of such techniques to other DL domains, such as the development of Graph Attention Networks [26,27,28,29]. Nevertheless, these networks have yet to find their way into the wind modelling community, as the authors of this paper cannot report on finding any research discussing the use of attention-based GNNs applied to wind energy.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Wind Park Power Prediction: Attention-Based Graph Networks and Deep Learning to Capture Wake Losses

Bentsen

Warakagoda

Stenbro

et al. 2022

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

With the increased penetration of wind energy into the power grid, it has become increasingly important to be able to predict the expected power production for larger wind farms. Deep learning (DL) models can learn complex patterns in the data and have found wide success in predicting wake losses and expected power production. This paper proposes a modular framework for attention-based graph neural networks (GNN), where attention can be applied to any desired component of a graph block. The results show that the model significantly outperforms a multilayer perceptron (MLP) and a bidirectional LSTM (BLSTM) model, while delivering performance on-par with a vanilla GNN model. Moreover, we argue that the proposed graph attention architecture can easily adapt to different applications by offering flexibility into the desired attention operations to be used, which might depend on the specific application. Through analysis of the attention weights, it was showed that employing attention-based GNNs can provide insights into what the models learn. In particular, the attention networks seemed to realise turbine dependencies that aligned with some physical intuition about wake losses.

show abstract

Section: Gnn Model With Attentionmentioning

confidence: 99%

Section: Gnn Model With Attentionmentioning

confidence: 99%

Section: Gnn Model With Attentionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Wind Park Power Prediction: Attention-Based Graph Networks and Deep Learning to Capture Wake Losses

Bentsen

Warakagoda

Stenbro

et al. 2022

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

show abstract

“…Chen & Chen [54] argue that different graphs may have different preferences for edges and weights and hence introduce Edge GATs. They extend the use of GATs to incorporate edge features in addition to the original node features.…”

Section: Edge Gats (Egats)mentioning

confidence: 99%

A Bird's-Eye Tutorial of Graph Attention Architectures

Dhole¹,

Yang²

2022

Preprint

View full text Add to dashboard Cite

Graph Neural Networks (GNNs) have shown tremendous strides in performance for graph-structured problems especially in the domains of natural language processing, computer vision and recommender systems. Inspired by the success of the transformer architecture, there has been an ever-growing body of work on attention variants of GNNs attempting to advance the state of the art in many of these problems. Incorporating "attention" into graph mining has been viewed as a way to overcome the noisiness, heterogenity and complexity associated with graph-structured data as well as to encode soft-inductive bias. It is hence crucial and advantageous to study these variants from a bird's-eye view to assess their strengths and weaknesses. We provide a systematic and focused tutorial centered around attention based GNNs in a hope to benefit researchers dealing with graph-structured problems. Our tutorial looks at GNN variants from the point of view of the attention function and iteratively builds the reader's understanding of different graph attention variants.

show abstract

Developing Relationships: A Heterogeneous Graph Network with Learnable Edge Representation for Emotion Identification in Conversations

Liang

et al. 2022

Artificial Intelligence

View full text Add to dashboard Cite

EGAT: Edge-Featured Graph Attention Network

Cited by 38 publications

References 4 publications

Wind Park Power Prediction: Attention-Based Graph Networks and Deep Learning to Capture Wake Losses

Wind Park Power Prediction: Attention-Based Graph Networks and Deep Learning to Capture Wake Losses

A Bird's-Eye Tutorial of Graph Attention Architectures

Developing Relationships: A Heterogeneous Graph Network with Learnable Edge Representation for Emotion Identification in Conversations

Contact Info

Product

Resources

About