Combining No-regret and Q-learning

Abstract

Counterfactual Regret Minimization (CFR) has found success in settings like poker which have both terminal states and perfect recall. We seek to understand how to relax these requirements. As a first step, we introduce a simple algorithm, local no-regret learning (LONR), which uses a Q-learning-like update rule to allow learning without terminal states or perfect recall. We prove its convergence for the basic case of MDPs (and limited extensions of them) and present empirical results showing that it achieves last iterate convergence in a number of settings, most notably NoSDE games, a class of Markov games specifically designed to be challenging to learn where no prior algorithm is known to achieve convergence to a stationary equilibrium even on average.

Generating intelligent agent behaviors in multi-agent game AI using deep reinforcement learning algorithm

“…In the agent's environment, the agent transitions to another state through the taken action. If the next state could be predicted without knowing/dependent on the preceded events, then the mathematical equation of the property is given in (1).…”

“…Reinforcement learning (RL), one of the machine learning (ML) methods, is utilized by these companies to train an expert agent who outperforms humans in the game. The utilization of the game in training the RL agent is to describe the complex and high-dimensional real-world data [1]- [9]. By utilizing games, RL researchers will be able to evade high experimental costs in training an agent to do intelligence tasks [10].…”