Reinforcement Learning CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   How can reinforcement learning be applied in AI testing?

   • ATo automatically generate test cases by learning which actions lead to system failures.
   • BTo reduce the need for test environments by using static code analysis.
   • CTo label test data manually for supervised learning models.
   • DTo replace all manual testing with automated regression testing.
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   Why is exploration important in reinforcement learning?

   • AExploration allows the agent to discover new actions that may lead to higher rewards than currently known.
   • BExploration is only needed in supervised learning, not reinforcement learning.
   • CExploration reduces the complexity of the learning algorithm.
   • DExploration guarantees that the agent will receive the maximum possible reward in every episode.
   Reveal Solution Discussion Select an answer first
4. 3expert · hard

   An RL agent is trained to control a traffic light. The reward is -1 for each car that waits more than 30 seconds. The agent learns to keep the light green for one direction almost always, causing long waits for the other direction. Which of the following is the most likely cause?

   • AThe value function is overestimating the value of keeping the light green.
   • BThe reward function does not penalize long waits for the other direction enough.
   • CThe state representation does not include the number of waiting cars.
   • DThe agent is exploring too much and not exploiting the learned policy.
   Reveal Solution Discussion Select an answer first
5. 4expert · hard

   In an RL system for a chatbot, the agent receives a reward only when the user gives positive feedback at the end of the conversation. The agent often fails to learn which specific responses led to the reward. Which of the following is the most effective way to address this credit assignment problem?

   • AIncrease the exploration rate to try more diverse responses.
   • BUse a supervised learning model to predict user feedback.
   • CUse a value function to estimate the expected future reward for each state-action pair.
   • DReduce the reward to a smaller value to make it less influential.
   Reveal Solution Discussion Select an answer first
6. 5expert · medium

   A team is considering using RL to optimize a manufacturing process. They have a simulator that accurately models the process. Which of the following is a key advantage of using a simulator over training directly on the real process?

   • AThe simulator eliminates the need for a reward function.
   • BThe simulator allows the agent to explore risky actions without real-world consequences.
   • CThe simulator reduces the need for exploration.
   • DThe simulator guarantees the learned policy will work in the real process.
   Reveal Solution Discussion Select an answer first