Math for ML quiz

Question 1

Why is math important for machine learning?

Accepted Answer

Statistics, linear algebra, calculus, and probability all underpin machine learning

Answer

Math is not needed; only coding skills matter for ML

Answer

Only basic addition and multiplication are used in ML

Answer

ML math is identical to basic school arithmetic

Question 2

What is the 'median' and when is it more useful than the mean?

Accepted Answer

The middle value when data is sorted, more resistant to outliers than the mean

Answer

The most frequently occurring value in the data

Answer

The average of the highest and lowest values

Answer

The value that appears exactly once in the dataset

Question 3

What is 'variance' in statistics?

Accepted Answer

A measure of how spread out values are around the mean

Answer

The difference between the highest and lowest values

Answer

The average squared error of a model

Answer

The number of unique values in a dataset

Question 4

What is 'probability' and how is it used in ML?

Accepted Answer

A number from 0 to 1 measuring how likely an event is

Answer

Only whole numbers between 0 and 100

Answer

The percentage of training examples used

Answer

A measure of how fast the model trains

Question 5

What is a 'matrix' and where does ML use matrices?

Accepted Answer

A 2D grid of numbers used to represent datasets and neural network weights

Answer

A type of AI model for image classification

Answer

A database structure for storing ML training data

Answer

A method for measuring model accuracy

Question 6

What is a 'derivative' and how does it relate to training ML models?

Accepted Answer

A measure of how much a function's output changes for a small input change, used to guide training

Answer

A measure of how fast the model can make predictions

Answer

The difference between predicted and actual values

Answer

The learning rate of a neural network

Question 7

What is 'learning rate' in gradient descent?

Accepted Answer

How large each adjustment step is during gradient descent training

Answer

The speed of the computer training the model

Answer

The number of examples processed per second

Answer

The rate at which a model generates predictions

Question 8

What is 'mean squared error' (MSE)?

Accepted Answer

The average squared difference between predicted and actual values

Answer

The average number of predictions per second

Answer

The percentage of wrong predictions

Answer

The difference between the largest and smallest errors

Question 9

What is the 'sigmoid function' and where is it used?

Accepted Answer

A function that squeezes any number to a value between 0 and 1 for probability output

Answer

A function that converts images to numerical values

Answer

A function that calculates the mean of a dataset

Answer

A function that generates random numbers

Question 10

What is 'Euclidean distance' and how is it used in ML?

Accepted Answer

The straight-line distance between two data points, used to find similar examples in algorithms like KNN

Answer

The distance between two database tables

Answer

The error distance between a prediction and the actual value

Answer

The time taken for data to travel between servers

Question 11

What does 'independent variable' vs 'dependent variable' mean in ML?

Accepted Answer

The independent variable is the input feature; the dependent variable is what you're predicting

Answer

The dependent variable is the input; the independent variable is the output

Answer

Both variables are always numerical

Answer

Only one variable can be used in any ML model

Question 12

What is a 'confusion matrix' in math terms?

Accepted Answer

A table counting correct and incorrect predictions for each class

Answer

A matrix that confuses the model during training

Answer

A chart that shows model accuracy over time

Answer

A matrix used only for regression problems

Question 13

What is 'Bayes' theorem' in simple terms?

Accepted Answer

A formula for updating probability beliefs when new evidence is received

Answer

A formula for calculating the mean of a dataset

Answer

A type of neural network layer

Answer

A method for sorting data alphabetically

Question 14

What is a 'hyperparameter' and how is it different from a regular parameter?

Accepted Answer

Hyperparameters are pre-training settings you choose; regular parameters are learned during training

Answer

Hyperparameters are the same as regular model parameters

Answer

Hyperparameters are only used in deep learning

Answer

Regular parameters are set manually; hyperparameters are learned automatically

Question 15

What does 'dimensionality' mean in data science?

Accepted Answer

The number of features (columns) in a dataset

Answer

The size in megabytes of a dataset

Answer

The number of rows in a dataset

Answer

The number of models in an ensemble

Question 16

What is 'correlation coefficient' and what does it tell you?

Accepted Answer

A number from -1 to 1 measuring the strength and direction of a linear relationship

Answer

A measure of how many features correlate with the target

Answer

A measure of model accuracy on the training set

Answer

The percentage of samples that are outliers

Question 17

What is 'softmax' function and where is it used?

Accepted Answer

A function that converts numbers into probabilities that sum to 1 for multiclass classification

Answer

A function that picks the maximum value in a list

Answer

A technique for resizing images before feeding them to a model

Answer

A method for combining multiple ML models

Question 18

What is 'exponential' growth and why does it matter in AI computing?

Accepted Answer

Growth that multiplies repeatedly, like doubling — 1, 2, 4, 8, 16

Answer

Growth that adds a fixed amount each time

Answer

Growth that eventually levels off to a maximum

Answer

Growth that decreases over time

Question 19

What is 'a^2 + b^2 = c^2' (Pythagorean theorem) doing in AI?

Accepted Answer

Used in KNN and k-means to calculate the distance between data points in multi-dimensional space

Answer

Only used for measuring physical distances in robotics

Answer

Only relevant for 2D datasets

Answer

A theorem only used in computer graphics, not ML

Question 20

What is 'the chain rule' and why does it power neural network training?

Accepted Answer

A calculus rule for derivatives of nested functions, enabling backpropagation in neural networks

Answer

A rule for chaining multiple ML models together

Answer

A rule that links a model's accuracy to its size

Answer

A method for connecting multiple datasets

Question 21

What is 'absolute value' and how does it appear in ML?

Accepted Answer

The positive version of a number (always non-negative), used in L1 loss calculations in ML

Answer

The most important value in a dataset

Answer

The average of positive and negative values

Answer

A value that is always exactly 1 in probability calculations

Question 22

What is 'modular arithmetic' and does it appear in AI?

Accepted Answer

Finding the remainder after division — appears in AI grokking research and hash functions

Answer

A type of machine learning for building modular code

Answer

A way of organizing neural network layers in modules

Answer

Arithmetic only used in cryptography, not AI

Question 23

What is 'matrix transpose' and why is it useful?

Accepted Answer

Flipping a matrix along its diagonal so rows become columns — used extensively in backpropagation

Answer

Moving a matrix from one variable to another

Answer

Multiplying all elements of a matrix by -1

Answer

Converting a matrix from integer to floating-point values

Question 24

What is 'the normal distribution' (bell curve) and its role in ML?

Accepted Answer

A symmetric bell-shaped distribution around a mean — used in weight initialization and noise modeling

Answer

A distribution describing only binary outcomes (0 or 1)

Answer

A distribution only found in financial data

Answer

A distribution with no mathematical formula

Question 25

What is 'matrix determinant' and does ML use it?

Accepted Answer

A number encoding whether a matrix can be inverted and how it scales volume — used in statistics and generative models

Answer

A measure of a matrix's size in megabytes

Answer

The maximum value found in a matrix

Answer

The number of unique values in a matrix

Question 26

What is 'polynomial features' and when does it help regression?

Accepted Answer

Adding higher powers of input features to let a linear model capture nonlinear (curved) relationships

Answer

Features that only include values between 0 and 1

Answer

Features that are computed using polynomials from two different datasets

Answer

A technique for reducing the number of features in a dataset

Test yourself