M2I2 improves multi-agent communication by adding masked state reconstruction and joint-action prediction auxiliary tasks, plus a meta-learned module that selects which observation dimensions to share.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.MA 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
M2I2: Learning Efficient Multi-Agent Communication via Masked State Modeling and Intention Inference
M2I2 improves multi-agent communication by adding masked state reconstruction and joint-action prediction auxiliary tasks, plus a meta-learned module that selects which observation dimensions to share.