Methods › General › Negative Sampling › MBS

Model-based Subsampling

MBS

11 papers tagged archive 2025-07-28

Introduced by Xincan Feng et al. in Model-based Subsampling for Knowledge Graph Completion

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

To avoid the problem caused by low-frequent entity-relation pairs, our MBS uses the estimated probabilities from a trained model θ′ to calculate frequencies for each triplet and query. By using θ′, the NS loss in KGE with MBS is represented as follows: ℓ_(mbs)(θ;θ′) = -1/(|D|)∑_((x,y) ∈D) [A_(mbs)(θ′)log(σ(s_θ(x,y)+γ)) +1/νsum_(yᵢ∼pₙ(yᵢ|x))^νB_(mbs)(θ′)log(σ(-s_θ(x,yᵢ)-γ))],

PaperSource

Papers archive 2025-07-28

11 shown of 11, 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

18 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
Super-Resolution3
Deep Learning1
Denoising1
Graph Embedding1
ISAC1
Integrated sensing and communication1
Knowledge Graph Completion1
Knowledge Graph Embedding1
Large Language Model1
Mamba1
Management1
Misinformation1
Model Compression1
Question Answering1
RAG1
Retrieval-augmented Generation1
Speech Enhancement1
model1

Usage over time archive 2025-07-28

Papers per year tagged with MBS: 2023 to 2025, peak 6 6 0 2023: 3 papers 2023 2024: 6 papers 2024 2025: 2 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (11 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

Negative Sampling

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