ML-Reinforcement Learning

Reinforcement Learning is a type of learning where an agent interacts with an environment and learns by:

• Imitating human behavior

• Observing unlabeled data

• Trial and error

• Calculating gradients

In Reinforcement Learning, what is the primary goal of the agent?

• To find the optimal policy

• To memorize the training data

• To perform classification tasks

• To minimize the error between predicted and actual values

Which of the following is NOT a component of a Reinforcement Learning system?

• Policy

• Environment

• Data Loader

• Reward Function

The process of the agent taking actions in the environment and receiving feedback is known as:

• Supervised Learning

• Unsupervised Learning

• Reinforcement Learning

• Semi-supervised Learning

What is the purpose of the reward signal in Reinforcement Learning?

• To define the agent's policy

• To measure the agent's performance

• To define the environment's dynamics

• To pre-train the agent's model

In Reinforcement Learning, what is an "episode"?

• A sequence of actions taken by the agent

• A set of training data samples

• A collection of states in the environment

• A complete interaction cycle from the initial state to the terminal state

Which algorithm is commonly used to make decisions and learn policies in Reinforcement Learning?

• K-Means

• Decision Trees

• Q-Learning

• Support Vector Machines

What does the Q-value represent in Q-Learning?

• The quality of the training data

• The probability of taking a certain action

• The expected cumulative reward for taking an action in a given state

• The percentage of correct predictions made by the agent

The trade-off between exploration and exploitation in Reinforcement Learning refers to:

• Balancing the number of training samples in each class

• Determining the learning rate for updating Q-values

• Choosing between different machine learning algorithms

• Deciding whether to take known actions or explore new actions in the environment

Which algorithm aims to learn the value function directly without explicitly representing the policy?

• Policy Gradient

• Deep Q Network (DQN)

• Monte Carlo Simulation

• Temporal Difference (TD) Learning