BEACON introduces several novel ideas to the design of implicit communication and efficient exploration.

ARacknowledgesfundingfroman Adobe Faculty Research Award, andan NSFDMS 1916320 grant.

We propose a novel approach to learning the generative neural fields represented by linear combinations of implicit basis networks.

Mirror descent [37] is becoming increasingly popular in a variety of settings in optimization and machine learning. One reason for its success is the fact that mirror descent can be adapted to fit the geometry ofthe optimization problem athand bychoosing asuitable strictly convexpotential function,theso-calledmirrormap.

Continual learning aims to learn a sequence of tasks from dynamic data distributions.

The proposed algorithm, aswell astheloss function, naturally generalize todifferent topological structures in both 2D and 3D settings.

Learning from AI feedback (LAIF) is a popular paradigm for improving the instruction-following abilities of powerful pre-trained language models.

Given 2 (0,12] and 2 (0,1), letT bethesamplesetgiventothe learner.