Stabilising experience replay for deep multi-agent reinforcement learning

J Foerster, N Nardelli, G Farquhar… - … machine learning, 2017 - proceedings.mlr.press
… This paper proposed two methods for stabilising experience replay in deep multi-agent
reinforcement learning: 1) using a multi-agent variant of importance sampling to naturally decay …

Stabilising Experience Replay for Deep MultiAgent Reinforcement Learning

S Whiteson - 2017 - cs.ox.ac.uk
… This paper proposed two methods for stabilising experience replay in deep multi-agent
reinforcement learning: 1) using a multi-agent variant of importance sampling to naturally decay …

Robust experience replay sampling for multi-agent reinforcement learning

IT Nicholaus, DK Kang - Pattern Recognition Letters, 2022 - Elsevier
… target specific experiences. These approaches try to stabilize experience replay, but in our
case, we are filtering samples to find which experiences are more suitable at particular states. …

Deep multi-agent reinforcement learning

J Foerster - 2018 - ora.ox.ac.uk
… for stabilising experience replay during centralised training using a version of multiagent
each agent during training. This metadata fingerprint disambiguates during which stage of …

Experience selection in multi-agent deep reinforcement learning

Y Wang, Z Zhang - 2019 IEEE 31st International Conference on …, 2019 - ieeexplore.ieee.org
replay drastically improves the utilization rate of experienceexperience replay with
multi-agent reinforcement learning is still an open challenge. In multi-agent reinforcement learning, …

Correcting experience replay for multi-agent communication

S Ahilan, P Dayan - arXiv preprint arXiv:2010.01192, 2020 - arxiv.org
… the problem of learning to communicate using multi-agent reinforcement learning (MARL). A
… sample from a multi-agent replay buffer which is used for off-policy learning. In general, the …

Experience augmentation: Boosting and accelerating off-policy multi-agent reinforcement learning

Z Ye, Y Chen, G Song, B Yang, S Fan - arXiv preprint arXiv:2005.09453, 2020 - arxiv.org
… the effectiveness of Experience Replay [13], reinforcement learning was suffered … experience
replay, it is typical to store the agent’s experience e = (o, a, r, o ) at each step into the replay

Lenient multi-agent deep reinforcement learning

G Palmer, K Tuyls, D Bloembergen… - arXiv preprint arXiv …, 2017 - arxiv.org
… Much of the success of single agent deep reinforcement learning (DRL) in recent years
can be attributed to the use of experience replay memories (ERM), which allow Deep Q-…

Deep reinforcement learning for multiagent systems: A review of challenges, solutions, and applications

TT Nguyen, ND Nguyen… - IEEE transactions on …, 2020 - ieeexplore.ieee.org
… two methods for stabilizing experience replay of DQN in MADRL… Gadi, “Multi-agent
reinforcement learning using linear fuzzy … Dusparic, “Multi-agent deep reinforcement learning for …

Learning from good trajectories in offline multi-agent reinforcement learning

Q Tian, K Kuang, F Liu, B Wang - … of the AAAI Conference on Artificial …, 2023 - ojs.aaai.org
… In online MARL, most works related to the experience replay focus on stable decentralized
multi-agent training (Foerster et al. 2017; Omidshafiei et al. 2017; Palmer et al. 2018), but …