Q#1: Complex decisions involve:
(A) Multiple sequential actions with uncertain outcomes
(B) Single-step deterministic choices
(C) BFS only
(D) DFS only
Answer: (A) Multiple sequential actions with uncertain outcomes
Q#2: Sequential decision problems are often modeled as:
(A) Markov Decision Processes (MDPs)
(B) Single-step decision networks
(C) BFS only
(D) DFS only
Answer: (A) Markov Decision Processes (MDPs)
Q#3: An MDP is defined by:
(A) States, actions, transition model, and reward function
(B) Single-step outcomes only
(C) BFS only
(D) DFS only
Answer: (A) States, actions, transition model, and reward function
Q#4: Transition model in MDPs defines:
(A) P(next state | current state, action)
(B) Reward only
(C) BFS only
(D) DFS only
Answer: (A) P(next state | current state, action)
Q#5: Reward function in MDPs specifies:
(A) Immediate payoff for each state or state-action pair
(B) Probability of transitions
(C) BFS only
(D) DFS only
Answer: (A) Immediate payoff for each state or state-action pair
Q#6: Policy in an MDP is:
(A) Mapping from states to actions
(B) Set of probabilities only
(C) BFS only
(D) DFS only
Answer: (A) Mapping from states to actions
Q#7: The goal in an MDP is to:
(A) Find a policy that maximizes expected cumulative reward
(B) BFS only
(C) DFS only
(D) Random action
Answer: (A) Find a policy that maximizes expected cumulative reward
Q#8: Discount factor in MDPs represents:
(A) Importance of future rewards
(B) Immediate reward only
(C) BFS only
(D) DFS only
Answer: (A) Importance of future rewards
Q#9: Value function in an MDP represents:
(A) Expected cumulative reward from a state
(B) Immediate probability only
(C) BFS only
(D) DFS only
Answer: (A) Expected cumulative reward from a state
Q#10: Optimal value function satisfies:
(A) Bellman equation
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Bellman equation
Q#11: Value iteration algorithm:
(A) Iteratively updates value function until convergence
(B) Computes policy randomly
(C) BFS only
(D) DFS only
Answer: (A) Iteratively updates value function until convergence
Q#12: Policy iteration algorithm:
(A) Alternates between policy evaluation and policy improvement
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Alternates between policy evaluation and policy improvement
Q#13: In partially observable MDPs (POMDPs), the agent:
(A) Cannot directly observe the current state
(B) Fully observes states
(C) BFS only
(D) DFS only
Answer: (A) Cannot directly observe the current state
Q#14: Belief state in POMDPs represents:
(A) Probability distribution over possible states
(B) Single deterministic state
(C) BFS only
(D) DFS only
Answer: (A) Probability distribution over possible states
Q#15: Solving POMDPs requires:
(A) Reasoning over belief states
(B) Deterministic reasoning only
(C) BFS only
(D) DFS only
Answer: (A) Reasoning over belief states
Q#16: Complex decisions often require:
(A) Multi-step planning under uncertainty
(B) Single-step expected utility only
(C) BFS only
(D) DFS only
Answer: (A) Multi-step planning under uncertainty
Q#17: Utility in complex decisions can be:
(A) Cumulative over multiple steps
(B) Only immediate reward
(C) BFS only
(D) DFS only
Answer: (A) Cumulative over multiple steps
Q#18: Monte Carlo methods in decision making are used to:
(A) Approximate expected rewards or value functions
(B) Compute exact values
(C) BFS only
(D) DFS only
Answer: (A) Approximate expected rewards or value functions
Q#19: Reinforcement learning is:
(A) Learning optimal policies through trial and error
(B) BFS only
(C) DFS only
(D) Deterministic planning
Answer: (A) Learning optimal policies through trial and error
Q#20: Q-learning is a method to:
(A) Learn optimal action-value function
(B) Compute exact value function analytically
(C) BFS only
(D) DFS only
Answer: (A) Learn optimal action-value function
Q#21: SARSA algorithm updates:
(A) Action-value based on current policy
(B) BFS only
(C) DFS only
(D) Random actions
Answer: (A) Action-value based on current policy
Q#22: Exploration in complex decision-making ensures:
(A) Trying actions to discover better rewards
(B) Only exploiting known actions
(C) BFS only
(D) DFS only
Answer: (A) Trying actions to discover better rewards
Q#23: Exploitation in complex decision-making is:
(A) Selecting best-known action
(B) Random exploration only
(C) BFS only
(D) DFS only
Answer: (A) Selecting best-known action
Q#24: Markov property assumes:
(A) Next state depends only on current state and action
(B) Full history of states
(C) BFS only
(D) DFS only
Answer: (A) Next state depends only on current state and action
Q#25: Complex decisions can be represented by:
(A) Decision trees, MDPs, or POMDPs
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Decision trees, MDPs, or POMDPs
Q#26: Dynamic programming in MDPs:
(A) Solves for optimal policies efficiently using Bellman equations
(B) BFS only
(C) DFS only
(D) Random actions
Answer: (A) Solves for optimal policies efficiently using Bellman equations
Q#27: Forward search in decision-making:
(A) Explores possible future states to evaluate actions
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Explores possible future states to evaluate actions
Q#28: Backward induction solves:
(A) Multi-step sequential decision problems
(B) Single-step decisions only
(C) BFS only
(D) DFS only
Answer: (A) Multi-step sequential decision problems
Q#29: Approximate solutions are needed when:
(A) State space is very large
(B) State space is small
(C) BFS only
(D) DFS only
Answer: (A) State space is very large
Q#30: Rollout algorithms:
(A) Evaluate actions using simulated future trajectories
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Evaluate actions using simulated future trajectories
Q#31: Stochastic policies:
(A) Assign probabilities to actions in each state
(B) Select a single deterministic action
(C) BFS only
(D) DFS only
Answer: (A) Assign probabilities to actions in each state
Q#32: Deterministic policies:
(A) Choose a single action for each state
(B) Randomize actions
(C) BFS only
(D) DFS only
Answer: (A) Choose a single action for each state
Q#33: Complex decisions often involve:
(A) Balancing exploration and exploitation
(B) BFS only
(C) DFS only
(D) Random choices
Answer: (A) Balancing exploration and exploitation
Q#34: Discounted reward ensures:
(A) Present rewards are valued more than distant future rewards
(B) All rewards are equal
(C) BFS only
(D) DFS only
Answer: (A) Present rewards are valued more than distant future rewards
Q#35: Policy evaluation computes:
(A) Value function for a given policy
(B) Best action only
(C) BFS only
(D) DFS only
Answer: (A) Value function for a given policy
Q#36: Policy improvement generates:
(A) Better policy based on value function
(B) BFS only
(C) DFS only
(D) Random policy
Answer: (A) Better policy based on value function
Q#37: Reinforcement learning agents:
(A) Learn from interaction with the environment
(B) BFS only
(C) DFS only
(D) Pre-programmed only
Answer: (A) Learn from interaction with the environment
Q#38: Temporal difference learning updates:
(A) Estimates using differences between successive predictions
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Estimates using differences between successive predictions
Q#39: Complex decision-making accounts for:
(A) Long-term consequences of actions
(B) BFS only
(C) DFS only
(D) Immediate outcomes only
Answer: (A) Long-term consequences of actions
Q#40: Model-based approaches in MDPs:
(A) Use known transition and reward functions
(B) Learn from trial and error only
(C) BFS only
(D) DFS only
Answer: (A) Use known transition and reward functions
Q#41: Model-free approaches in MDPs:
(A) Learn value function or policy directly from experience
(B) Use full transition model
(C) BFS only
(D) DFS only
Answer: (A) Learn value function or policy directly from experience
Q#42: Simulation is useful in complex decisions for:
(A) Evaluating policies without explicit computation
(B) BFS only
(C) DFS only
(D) Deterministic reasoning only
Answer: (A) Evaluating policies without explicit computation
Q#43: Complex decision-making requires:
(A) Handling uncertainty, sequential actions, and long-term rewards
(B) Single-step decisions only
(C) BFS only
(D) DFS only
Answer: (A) Handling uncertainty, sequential actions, and long-term rewards
Q#44: Approximate dynamic programming helps:
(A) Solve large MDPs efficiently
(B) BFS only
(C) DFS only
(D) Deterministic planning
Answer: (A) Solve large MDPs efficiently
Q#45: Exploration in reinforcement learning prevents:
(A) Converging to suboptimal policies
(B) BFS only
(C) DFS only
(D) Random action
Answer: (A) Converging to suboptimal policies
Q#46: Exploitation ensures:
(A) Using the best-known actions to gain rewards
(B) BFS only
(C) DFS only
(D) Random action
Answer: (A) Using the best-known actions to gain rewards
Q#47: Multi-step planning considers:
(A) Future states and cumulative rewards
(B) Immediate reward only
(C) BFS only
(D) DFS only
Answer: (A) Future states and cumulative rewards
Q#48: Bellman optimality equation defines:
(A) Relationship between value of a state and values of successor states
(B) BFS only
(C) DFS only
(D) Random assignments
Answer: (A) Relationship between value of a state and values of successor states
Q#49: Complex decisions in AI combine:
(A) Probabilistic reasoning, sequential planning, and reward optimization
(B) BFS only
(C) DFS only
(D) Deterministic one-step choice
Answer: (A) Probabilistic reasoning, sequential planning, and reward optimization
Q#50: The main goal of making complex decisions in AI is:
(A) Find policies that maximize long-term expected rewards under uncertainty
(B) BFS only
(C) DFS only
(D) Random action
Answer: (A) Find policies that maximize long-term expected rewards under uncertainty