419.419. Prioritized Experience Replay

Prioritized Experience Replay

=====================================

In this problem, you will implement Prioritized Experience Replay (PER), a technique used in Deep Reinforcement Learning to improve the stability and efficiency of training agents.

**Example:*

Suppose we have a simple Markov Decision Process (MDP) with two states (A and B) and two actions (left and right). The agent starts at state A and receives a reward of -1 for each step. The goal is to reach state B.

| State | Action | Next State | Reward |

| --- | --- | --- | --- |

| A | left | A | -1 |

| A | right | B | 10 |

| B | left | B | -1 |

| B | right | A | -1 |

The agent starts at state A and chooses the right action, reaching state B. The reward is 10, which is higher than the expected reward of -1. This experience should be prioritized in the replay buffer.

Constraints:

**Note:** This problem assumes you have a basic understanding of Deep Reinforcement Learning and the concept of experience replay.