The SP algorithm requires the voters to predict other voters' report in the form of a full probability distribution over all rankings of alternatives.

Is the sequential or alternating training scheme better (R4)? Comparisontoautoencoders(AEs),VAEs,GANs(R1,R2,R3).Weagreethatacomparisonto(V)AEsandGANs27 on generative and dimensionality reduction tasks would be very interesting. In the particle physics task, it seems unfair to compare on a new metric of inference of underlying parame-36 ters(R1). Likelihood-free inference(LFI) is its own thriving research area with applications from neuroscience to38 epidemiology.

It communicates among agents within the system to collaboratively solve tasks, under the premise of shared information. However, when agents' collaborations are leveraged to perform multi-person tasks, a new challenge arisesduetoinformation asymmetry,sinceeachagentcanonlyaccess theinformationofitshumanuser.