Supervised learning

Question 1

Q: What does supervised learning refer to?

Answer 1

A: Algorithms that learn input (x) to output (y) mappings by being given examples to learn from, including the correct labels (y) for given inputs (x).

Question 2

Q: How does a supervised learning algorithm learn?

Answer 2

A: By seeing correct pairs of input (x) and desired output label (y), the algorithm learns to predict the output (y) given new input (x).

Question 3

Q: Give an example of supervised learning in email filtering.

Answer 3

A: Input: Email, Output: Spam or Not Spam (Spam Filter).

Question 4

Q: Provide an example of supervised learning in speech recognition.

Answer 4

A: Input: Audio Clip, Output: Text Transcript.

Question 5

Q: How is supervised learning used in machine translation?

Answer 5

A: Input: Text in one language (e.g., English), Output: Text in another language (e.g., Spanish).

Question 6

Q: Describe the use of supervised learning in online advertising.

Answer 6

A: Input: Information about an ad and user, Output: Prediction of whether the user will click on the ad or not, which drives revenue for ad platforms.

Question 7

Q: What is the role of supervised learning in self-driving cars?

Answer 7

A: Input: Image and sensor data, Output: Position of other cars to navigate safely.

Question 8

Q: Explain how supervised learning is used in manufacturing.

Answer 8

A: Input: Picture of a manufactured product, Output: Detection of defects like scratches or dents (Visual Inspection).

Question 9

Q: How does the housing price prediction example illustrate supervised learning?

Answer 9

A: By providing the learning algorithm with data pairs of house sizes and prices to predict the price of a new house based on its size.

Question 10

Q: What is regression in the context of supervised learning?

Answer 10

A: Predicting a continuous number, such as house prices, from infinitely many possible values.

Question 11

Q: What is the difference between fitting a straight line and a curve in supervised learning?

Answer 11

A: Fitting a straight line is a simpler model, whereas fitting a curve can capture more complex relationships between input (x) and output (y).

Question 12

Q: What is the other major type of supervised learning problem apart from regression?

Answer 12

A: Classification, which involves predicting discrete labels such as spam or not spam.

Question 13

Q: Why is supervised learning called “supervised”?

Answer 13

A: Because the learning algorithm is provided with a dataset that includes the correct labels for the input data, guiding the learning process.

Question 14

Q: What do supervised learning algorithms learn to predict?

Answer 14

A: Input to output mappings (X to Y).

Question 15

Q: What is a regression algorithm in supervised learning?

Answer 15

A: A type of supervised learning algorithm that predicts numbers from infinitely many possible numbers.

Question 16

Q: What is a classification algorithm in supervised learning?

Answer 16

A: A type of supervised learning algorithm that predicts categories, which are a limited set of possible outputs.

Question 17

Q: Give an example of a classification problem.

Answer 17

A: Breast cancer detection, where the goal is to classify a tumor as benign (0) or malignant (1).

Question 18

Q: How can classification be visualized when predicting tumor malignancy?

Answer 18

A: By plotting the size of the tumor on the horizontal axis and using 0 or 1 on the vertical axis to represent benign or malignant tumors.

Question 19

Q: What distinguishes classification from regression?

Answer 19

A: Classification predicts a small, finite set of possible categories, while regression predicts any number from an infinite set.

Question 20

Q: Can classification problems have more than two categories?

Answer 20

A: Yes, for example, a cancer diagnosis could predict multiple types of cancer, resulting in categories like 0, 1, or 2.

Question 21

Q: What terms are often used interchangeably in classification?

Answer 21

A: Output classes and output categories.

Question 22

Q: How does classification handle non-numeric categories?

Answer 22

A: It can predict categories such as whether a picture is of a cat or a dog.

Question 23

Q: How does the learning algorithm use multiple inputs for classification?

Answer 23

A: By finding a boundary that separates the different categories (e.g., benign vs. malignant tumors) based on multiple input features.

Question 24

Q: What is the significance of the boundary line found by the learning algorithm in classification?

Answer 24

A: It helps classify new inputs, aiding in diagnosis or prediction.