Methods › Computer Vision › Rendezvous › MHMA

Multi-Heads of Mixed Attention

MHMA

1 paper tagged archive 2025-07-28

Introduced by Chinedu Innocent Nwoye et al. in Rendezvous: Attention Mechanisms for the Recognition of Surgical Action Triplets in Endoscopic Videos

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

The multi-head of mixed attention combines both self- and cross-attentions, encouraging high-level learning of interactions between entities captured in the various attention features. It is build with several attention heads, each of the head can implement either self or cross attention. A self attention is when the key and query features are the same or come from the same domain features. A cross attention is when the key and query features are generated from different features. Modeling MHMA allows a model to identity the relationship between features of different domains. This is very useful in tasks involving relationship modeling such as human-object interaction, tool-tissue interaction, man-machine interaction, human-computer interface, etc.

PaperSourceSee Code · CAMMA-public/rendezvous

Papers archive 2025-07-28

1 shown of 1, 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

2 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
Action Triplet Recognition1
Triplet1

Usage over time archive 2025-07-28

Papers per year tagged with MHMA: 2021 to 2021, peak 1 1 0 2021: 1 paper 2021
Papers per year the archive tags with this method, by the paper's archive date (1 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

RendezvousVision TransformersTransformersAttention MechanismsAttentionAttention 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