Lenience towards Teammates Helps in Cooperative Multiagent Learning

20 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Negative Update Intervals in Deep Multi-Agent Reinforcement Learning (2018, background): “This can lead to action shadowing, where miscoordination due to sub-optimal joint-policies results in utility values of optimal actions being underestimated [5, 20].”
A survey and critique of multiagent deep reinforcement learning (2018, background): “The purpose of leniency is to overcome a pathology called relative overgeneralization [34, 223, 224].”
Theoretical Advantages of Lenient Learners: An Evolutionary Game Theoretic Perspective (2008, methodology): “In previous work, we have demonstrated the efficacy of algorithms based on this notion: specifically, the Lenient Multiagent Q-learning algorithm (Panait et al., 2006) and the iCCEA algorithm (Panait and Luke, 2006).”
Lenient Learning in Independent-Learner Stochastic Cooperative Games (2016, methodology): “In (Panait et al., 2006b, 2013) we demonstrated a lenient learning algorithm, LMRL, for repeated games with independent learners, comparing it favorably to the FMQ algorithm (Kapetanakis and Kudenko, 2002) for variations of the Climb and Penalty games (Claus and Boutilier, 1998).”
A Survey of Learning in Multiagent Environments: Dealing with Non-Stationarity (2017, background): “This extension incorporates the concept of leniency (Panait et al., 2006) to account for initial mis-coordination, which enables LFAQ to obtain high convergence to Pareto optimal equilibria in cooperative games.”