Reinforcement Learning CT-AI Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   An RL agent is trained to play a board game. The agent has a policy that always chooses the move with the highest estimated value. However, it loses to a simpler heuristic opponent. Which of the following is the most likely reason?

   • AThe reward function is too sparse to learn anything.
   • BThe value function is overestimating the value of losing moves.
   • CThe agent is exploring too much, causing it to make random moves.
   • DThe agent is not exploring enough, so its value estimates are inaccurate.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   What is the role of the environment in a reinforcement learning system?

   • AThe environment provides the agent with the current state and returns a reward and next state after each action.
   • BThe environment defines the learning algorithm used by the agent.
   • CThe environment is the policy that maps states to actions.
   • DThe environment stores the agent's past experiences and memories.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A team is building an AI system to automatically adjust the difficulty of a game in real time based on player performance. The system receives a score after each level and must learn which difficulty setting keeps players engaged. Which learning approach is most appropriate, and why?

   • ASelf-supervised learning, because the system can predict the next difficulty level from past player actions.
   • BUnsupervised learning, because the system can cluster players by skill level without needing labels.
   • CReinforcement learning, because the system can learn a policy by maximizing cumulative rewards from player engagement signals.
   • DSupervised learning, because the system can be trained on labeled examples of player engagement.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   How do actions and rewards interact in the reinforcement learning process?

   • ARewards are determined solely by the agent's policy, independent of the environment.
   • BThe agent takes an action, and the environment responds with a reward and a new state.
   • CActions are chosen by the environment, and rewards are chosen by the agent.
   • DActions and rewards are independent; actions do not influence future rewards.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A company wants to use reinforcement learning to optimize the order in which a robot picks items in a warehouse. Which of the following is a key advantage of RL over a rule-based approach in this scenario?

   • ARL can adapt to changing warehouse layouts and item availability without manual reprogramming.
   • BRL guarantees the optimal picking order in all cases.
   • CRL does not require a reward function.
   • DRL is easier to debug than rule-based systems.
   Reveal Solution Discussion Select an answer first