Define the Action-Selection Policy for an RL Agent — rl_define_policy • rlsims

Define the Action-Selection Policy for an RL Agent

rl_define_policy(policy, ...)

Arguments

policy: What policy should a decision be made under? Currently supported are softmax, greedy, and epsilon-greedy.
...: Additional arguments passed to or from specific methods, such as tau when policy = "softmax" and epsilon when policy = "epsilonGreedy".

Value

The agent object with modified policy structure.

Examples


# Example arguments include:

# (policy = "softmax", tau = 7)

# (policy = "epsilonGreedy", epsilon = 0.4)

# (policy = "greedy")