Socially Supervised Representation Learning: the Role of Subjectivity in Learning Efficient Representations

no code implementations • 20 Sep 2021 • Julius Taylor, Eleni Nisioti, Clément Moulin-Frier

In this work, we propose that aligning internal subjective representations, which naturally arise in a multi-agent setup where agents receive partial observations of the same underlying environmental state, can lead to more data-efficient representations.

Representation Learning Self-Supervised Learning