In the context of the multi-armed bandit problem, what is regret?
- The feeling of loss and remorse
- An optimization metric
- A random variable
- An arm selection policy
In the context of the multi-armed bandit problem, regret is an optimization metric that quantifies how much an agent's total reward falls short of the best possible reward it could have achieved by always choosing the best arm. It's a way to measure how well an agent's arm selection policy performs.
When NLP systems try to understand the context of words in medical documents to extract meaningful information, they are leveraging a technique called ________.
- Named Entity Recognition
- Sentiment Analysis
- Document Summarization
- Word Embeddings
Named Entity Recognition is a technique in NLP used to identify and classify entities in medical documents, such as drugs, diseases, or patient names.
In which scenario is unsupervised learning least suitable: predicting house prices based on features, grouping customers into segments, or classifying emails as spam or not spam?
- Classifying emails as spam or not spam
- Grouping customers into segments
- Predicting house prices based on features
- Unsupervised learning is suitable for all scenarios
Unsupervised learning is least suitable for classifying emails as spam or not spam. This is because unsupervised learning doesn't have labeled data to distinguish between spam and non-spam emails. It is more applicable to clustering or grouping data when you don't have clear labels, such as grouping customers into segments.
For a non-linearly separable dataset, which property of SVMs allows them to classify the data?
- Feature selection
- Kernel functions
- Large training dataset
- Parallel processing
SVMs can classify non-linearly separable data using kernel functions, which map the data into a higher-dimensional space where it becomes linearly separable.
In the context of reinforcement learning, what does the term "gradient" in "Policy Gradient Methods" refer to?
- The direction of steepest ascent in the policy space
- A mathematical term used to describe the rate of change
- The probability distribution of actions
- The value function
In "Policy Gradient Methods," the "gradient" refers to the direction in the policy space that increases the expected reward. It guides policy updates to maximize reward.
Which regression method assumes a linear relationship between the independent and dependent variables?
- Decision Tree Regression
- Logistic Regression
- Polynomial Regression
- Ridge Regression
Polynomial Regression assumes a linear relationship between the independent and dependent variables. It models relationships as polynomial functions, and other regression methods may assume different relationships.
A medical research company is working on image data, where they want to classify microscopic images into cancerous and non-cancerous categories. The boundary between these categories is not linear. Which algorithm would be a strong candidate for this problem?
- Convolutional Neural Network (CNN)
- Logistic Regression
- Naive Bayes Classifier
- Principal Component Analysis
Convolutional Neural Networks (CNNs) are excellent for image classification tasks, especially when dealing with non-linear boundaries. They use convolutional layers to extract features from images, making them suitable for tasks like cancerous/non-cancerous image classification.
The term "exploitation" in reinforcement learning refers to which of the following?
- Utilizing the best-known actions
- Trying new, unexplored actions
- Maximizing exploration
- Modifying the environment
Exploitation involves utilizing the best-known actions to maximize rewards based on current knowledge, minimizing risk and uncertainty.
________ learning is often used for discovering hidden patterns in data.
- Reinforcement
- Semi-supervised
- Supervised
- Unsupervised
Unsupervised learning is a machine learning approach where algorithms are used to identify patterns in data without explicit guidance. It is commonly employed for data exploration and pattern discovery.
When dealing with high-dimensional data, which of the two algorithms (k-NN or Naive Bayes) is likely to be more efficient in terms of computational time?
- Both Equally Efficient
- Naive Bayes
- Neither is Efficient
- k-NN
Naive Bayes is typically more efficient in high-dimensional data due to its simple probabilistic calculations, while k-NN can suffer from the "curse of dimensionality."