</> Code with Papers by Syntology

Methods › General › Attention Modules › Graph Self-Attention

Graph Self-Attention

64 papers tagged archive 2025-07-28

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

Graph Self-Attention (GSA) is a self-attention module used in the BP-Transformer architecture, and is based on the graph attentional layer.

For a given node u, we update its representation according to its neighbour nodes, formulated as 𝐡ᵤ ←GSA(𝒢, 𝐡ᵘ).

Let 𝐀(u) denote the set of the neighbour nodes of u in 𝒢, GSA(𝒢, 𝐡ᵘ) is detailed as follows:

𝐀ᵘ = concat({𝐡ᵥ | v ∈𝒜(u)})

𝐐ᵘᵢ = 𝐇ₖ𝐖^Qᵢ,𝐊ᵢᵘ = 𝐀ᵘ𝐖ᴷᵢ,𝐕ᵘᵢ = 𝐀ᵘ𝐖ᵢⱽ

headᵘᵢ = softmax(𝐐ᵘᵢ𝐊ᵢᵘᵀ/(√(d)))𝐕ᵢᵘ

GSA(𝒢, 𝐡ᵘ) = [headᵘ₁, …, headᵘₕ]𝐖ᴼ

where d is the dimension of h, and 𝐖^Qᵢ, 𝐖ᴷᵢ and 𝐖ⱽᵢ are trainable parameters of the i-th attention head.

Source: BP-Transformer: Modelling Long-Range Context via Binary...See Code · yzh119/BPT

Papers archive 2025-07-28

30 shown of 64, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.

Tasks archive 2025-07-28

20 shown of 86 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Graph Attention3
Graph Classification3
Representation Learning3
Sentence3
Translation3
Abstractive Dialogue Summarization2
Anomaly Detection2
Articles2
Autonomous Driving2
BIG-bench Machine Learning2
Graph Regression2
Graph Representation Learning2
Language Modeling2
Language Modelling2
Machine Translation2
Molecular Property Prediction2
Sentiment Analysis2
Aspect-Based Sentiment Analysis (ABSA)1
Attribute1
Autonomous Vehicles1

Usage over time archive 2025-07-28

Papers per year tagged with Graph Self-Attention: 2019 to 2025, peak 17 17 0 2019: 1 paper 2019 2020: 4 papers 2020 2021: 9 papers 2021 2022: 10 papers 2022 2023: 17 papers 2023 2024: 13 papers 2024 2025: 10 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (64 dated). Bars are counts, not a trend claim.

Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).

Categories archive 2025-07-28

Attention Modules

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections