Crash Course Computer Science: Machine Learning and AI

What is the primary goal of machine learning?

• To make decisions based on data

• To organize data efficiently

• To store large amounts of data

• To display data in a user-friendly way

Which of the following is NOT an example of a question that machine learning can help answer?

• What video should YouTube recommend next?

• Is an email spam?

• Does a person’s heart have arrhythmia?

• How to cook a specific recipe?

In the context of machine learning, what is a 'classifier'?

• A way to display data

• A method for storing data

• An algorithm that makes decisions about data

• A type of computer hardware

What are 'features' in machine learning?

• Values that characterize the things we wish to classify

• Labels given to data

• Types of machine learning models

• Algorithms used for classification

What is 'labeled data'?

• Data that has been processed by an algorithm

• Data that has been visualized in a scatterplot

• Data that has been tagged with the correct answer

• Data that has been organized into categories

What is a 'confusion matrix' used for?

• To organize features

• To show where a classifier gets things right and wrong

• To visualize decision boundaries

• To store large datasets

What is the main job of machine learning algorithms?

• To maximize correct classifications while minimizing errors

• To store data efficiently

• To display data in a user-friendly way

• To organize data into categories

What is a 'decision tree' in machine learning?

• A type of data storage

• A method for visualizing data

• An algorithm that divides decision space into boxes

• A way to label data

What is a 'Support Vector Machine'?

• A type of decision tree

• A method for storing data

• An algorithm that slices up decision space using arbitrary lines

• A way to visualize data

What is the challenge with real-world machine learning problems?

• They often involve many features, making visualization difficult

• They are easy to solve by hand

• They require a lot of labeled data

• They often have too few features