As a challenging multi-player card game, DouDizhu has recently drawn much attention for analyzing competition and collaboration in imperfect-information games. In this paper, we propose PerfectDou, a state-of-the-art DouDizhu AI system that dominates the game, in an actor-critic framework with a proposed technique named perfect information distillation.

We study the embedding dimension of distance comparison data in two settings: contrastive learning and k -nearest neighbors ( k -NN).

We compare the algorithms on a wide range of large, convolutional and transformer-based neural network architectures.

Unlikeknowledge-based anddata-basedalgorithms, competence-based algorithms simultaneously address both the exploration challenge as well as distilling the generated experience in the form of reusable skills.

There are still, however, some unsolved challenges for a successful deployment of RL such as efficient learning of constrained or safe Markov Decision Processes (MDPs) [4].

Code is available at https://github.com/zfkarl/DANCE.

Theperformance ofneural networksonhigh-dimensional datadistributions suggests that it may be possible to parameterize a representation of agiven highdimensional function with controllably small errors, potentially outperforming standard interpolation methods. We demonstrate, both theoretically and numerically, that this is indeed the case. We map the parameters of a neural network to a system of particles relaxing with an interaction potential determined by the lossfunction.

Images taken between 2013 and 2015 are used as the validation set.

Neural Information Processing Systems http://nips.cc/