selectEpsilonGreedyAction {ReinforcementLearning}

Performs \varepsilon-greedy action selection

Description

Implements \varepsilon-greedy action selection. In this strategy, the agent explores the environment by selecting an action at random with probability \varepsilon. Alternatively, the agent exploits its current knowledge by choosing the optimal action with probability 1-\varepsilon.

Usage

selectEpsilonGreedyAction(Q, state, epsilon)

Arguments

Q

State-action table of type hash.

state

The current state.

epsilon

Exploration rate between 0 and 1.

Value

Character value defining the next action.

References

Sutton and Barto (1998). "Reinforcement Learning: An Introduction", MIT Press, Cambridge, MA.