Reinforcement Learning
التعلم المعزز (RL)
Definition
A type of machine learning in which an agent learns to make decisions through trial and interaction with an environment so as to maximize the total reward.
Definition
Reinforcement learning is a paradigm of machine learning in which an agent learns to choose actions through trial-and-error interaction with an environment so as to maximise cumulative reward.
Practical examples
Q = {("s", "a"): 2.0, ("s2", "a"): 5.0, ("s2", "b"): 3.0}
alpha, gamma, r = 0.5, 0.9, 1.0
best_next = max(Q[("s2", x)] for x in ["a", "b"])
Q[("s", "a")] += alpha * (r + gamma * best_next - Q[("s", "a")])
print(Q[("s", "a")])Expected output
3.75
Mathematical equations
Q(s,a) \leftarrow Q(s,a) + \alpha \big[\, r + \gamma \max_{a'} Q(s',a') - Q(s,a) \,\big]Scientific sources
References and scientific sources
-
BookReinforcement Learning: An Introduction (2nd ed.) The standard reference on reinforcement learning, freely available.
-
Research paperHuman-level control through deep reinforcement learning The DQN network that played Atari games.
