wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

Machine Learning Concepts Worksheet

Total questions: 60

Worksheet time: 33mins

Name
Class
Date
1.

Which of the following is a supervised learning algorithm?

a)

K-Means

b)

Apriori

c)

Decision Tree

d)

PCA

2.

What is the main goal of unsupervised learning?

a)

Predict labels

b)

Reduce dimensionality

c)

Find patterns in data

d)

Minimize loss

3.

Which technique is used to prevent overfitting in machine learning?

a)

A) Cross-validation

b)

B) Gradient boosting

c)

C) Early initialization

d)

D) Grid sampling

4.

Which algorithm is best suited for solving a regression problem?

a)

Logistic Regression

b)

Decision Tree Regressor

c)

Naive Bayes

d)

K-Means

5.

Which of the following is not a distance metric?

a)

Manhattan

b)

Euclidean

c)

Cosine

d)

Entropy

6.

Which concept explains the tradeoff between bias and variance?

a)

Gradient descent

b)

Regularization

c)

Bias-Variance Tradeoff

d)

Dimensionality reduction

7.

Which of these is not a type of machine learning?

a)

Reinforcement

b)

Supervised

c)

Heuristic

d)

Unsupervised

8.

Which ML technique would be most appropriate for detecting credit card fraud?

a)

Regression

b)

Classification

9.

Which method is used for feature selection?

a)

Dropout

b)

Random Forest Feature Importance

c)

Bagging

d)

Pooling

10.

Which activation function is used in binary classification?

a)

ReLU

b)

Tanh

c)

Softmax

d)

Sigmoid

11.

Which library is mainly used for data manipulation in Python?

a)

Matplotlib

b)

Pandas

c)

NumPy

d)

Seaborn

12.

Gradient Descent minimizes which of the following?

a)

Accuracy

b)

Precision

c)

Cost Function

d)

Recall

13.

In random forest, how is randomness introduced?

a)

Randomly selecting data samples and features

b)

Random activation functions

c)

Random weights

d)

Random output threshold

14.

What is the vanishing gradient problem?

a)

Gradients explode to infinity

b)

Gradients become too small to update weights

c)

Weights vanish

d)

Model accuracy drops suddenly.

15.

Which of the following is not a hyperparameter in a decision tree?

a)

Max depth

b)

Min samples split

c)

Learning rate

d)

Criterion

16.

Which ML technique is used in recommendation systems?

a)

Classification

b)

Clustering

c)

Collaborative Filtering

d)

Regression

17.

The ROC curve is plotted with:

a)

Precision vs Recall

b)

TPR vs FPR

c)

Accuracy vs Precision

d)

F1 vs Recall

18.

Which library in Python is used for deep learning?

a)

NumPy

b)

Matplotlib

c)

TensorFlow

d)

Pandas

19.

Which evaluation metric measures the area under the ROC curve?

a)

Precision

b)

Recall

c)

AUC

d)

F1-score

20.

What is the function of the loss function in training a model?

a)

Improve model accuracy

b)

Measure prediction error

c)

Stop overfitting

d)

Tune hyperparameters

21.

Which is true about the Naive Bayes algorithm?

a)

Assumes features are dependent

b)

Requires normalization

c)

Works poorly with categorical data

d)

Assumes conditional independence

22.

Which ML algorithm works on the principle of majority voting?

a)

Logistic Regression

b)

KNN

c)

SVM

d)

Naive Bayes

23.

Which dimensionality reduction technique is linear?

a)

t-SNE

b)

PCA

c)

UMAP

d)

Autoencoder

24.

A ride-sharing app wants to estimate the arrival time of a vehicle. Which type of ML problem is this?

a)

Classification

b)

Reinforcement Learning

c)

Regression

d)

Clustering

25.

Which AI technique powers voice assistants like Siri or Alexa?

a)

Clustering

b)

Natural Language Processing (NLP)

c)

Reinforcement Learning

d)

Computer Vision

26.

Which algorithm would be best for customer segmentation in marketing?

a)

Decision Tree

b)

K-Means

c)

Linear Regression

d)

Naive Bayes

27.

In email spam detection, what kind of learning problem is being solved?

a)

Clustering

b)

Regression

c)

Classification

d)

Reinforcement Learning

28.

An e-commerce website recommends products based on user behavior. What is this an example of?

a)

Classification

b)

Regression

c)

Collaborative Filtering

d)

Clustering

29.

Which technique would be most suitable for detecting anomalies in network traffic?

a)

Clustering

b)

Regression

c)

Classification

d)

Supervised Learning

30.

In facial recognition, what type of AI is being used?

a)

NLP

b)

Computer Vision

c)

Reinforcement Learning

d)

Genetic Algorithm

31.

Which method would be used to forecast stock prices?

a)

Classification

b)

Regression

c)

Clustering

d)

Reinforcement Learning

32.

A self-driving car deciding to stop or go is an example of:

a)

Supervised Learning

b)

Classification

c)

Reinforcement Learning

d)

Clustering

33.

Which ML technique is typically used in optical character recognition (OCR)?

a)

Clustering

b)

CNN

c)

RNN

d)

SVM

34.

Which of the following techniques can be used to handle vanishing gradients in deep neural networks?

a)

Batch Normalization

b)

Weight Initialization

c)

ReLU Activation

d)

Max Pooling

35.

What is the main advantage of using residual connections in ResNets?

a)

Reduces memory usage

b)

Avoids overfitting

c)

Enables training of deeper networks

d)

Reduces number of parameters

36.

In a Transformer model, which component is responsible for modeling sequence order?

a)

Positional Encoding

b)

Multi-head Attention

c)

Feedforward Layer

d)

Layer Normalization

37.

Which reinforcement learning algorithm uses both a policy and a value function?

a)

Q-Learning

b)

SARSA

c)

Deep Q Network

d)

Actor-Critic

38.

Which of the following is true about LSTM networks?

4 lines
39.

What is the key idea behind attention mechanisms in NLP models?

a)

Compress sequences

b)

Focus on important parts of the input

c)

Encode sequence length

d)

Reduce overfitting

40.

Which algorithm is most suitable for solving multi-armed bandit problems?

a)

Q-Learning

b)

Softmax Classifier

c)

Upper Confidence Bound (UCB)

d)

Monte Carlo Tree Search

41.

In Variational Autoencoders (VAEs), what is the role of the KL divergence term in the loss function?

a)

Encourage sparsity

b)

Ensure reconstruction accuracy

c)

Regularize latent space distribution

d)

Minimize overfitting

42.

What differentiates Gradient Boosting from Bagging techniques like Random Forests?

a)

Use of decision trees

b)

Parallel training

c)

Sequential model training using residuals

d)

Sampling with replacement

43.

Which property does the softmax function NOT satisfy?

a)

Outputs are between 0 and 1

b)

Outputs sum to 1

c)

Preserves order of inputs

d)

Used in multi-class classification

44.

What is the role of the “critic” in Actor-Critic RL methods?

a)

Suggests next action

b)

Estimates value function

c)

Optimizes the environment

d)

Calculates policy gradient

45.

Which concept is central to the success of self-supervised learning?

a)

Data Augmentation

b)

Ensemble Learning

46.

Which of the following methods is used to interpret predictions from black-box models?

a)

Logistic Regression

b)

LIME

c)

Grid Search

d)

Q-Learning

47.

Which of the following techniques can be used to handle vanishing gradients in deep neural networks?

a)

Batch Normalization

b)

Weight Initialization

c)

ReLU Activation

d)

Max Pooling

48.

What is the main advantage of using residual connections in ResNets?

a)

Reduces memory usage

b)

Avoids overfitting

c)

Enables training of deeper networks

d)

Reduces number of parameters

49.

In a Transformer model, which component is responsible for modeling sequence order?

a)

Positional Encoding

b)

Multi-head Attention

c)

Feedforward Layer

d)

Layer Normalization

50.

Which reinforcement learning algorithm uses both a policy and a value function?

a)

Q-Learning

b)

SARSA

c)

Deep Q Network

d)

Actor-Critic

51.

Which of the following is true about LSTM networks?

a)

They can only handle fixed-length sequences

b)

They suffer from vanishing gradients more than RNNs

c)

They use gating mechanisms to control memory flow

d)

They require convolutional layers to function

52.

What is the key idea behind attention mechanisms in NLP models?

a)

Compress sequences

b)

Focus on important parts of the input

c)

Encode sequence length

53.

Which algorithm is most suitable for solving multi-armed bandit problems?

a)

Q-Learning

b)

Softmax Classifier

c)

Upper Confidence Bound (UCB)

d)

Monte Carlo Tree Search

54.

In Variational Autoencoders (VAEs), what is the role of the KL divergence term in the loss function?

a)

Encourage sparsity

b)

Ensure reconstruction accuracy

c)

Regularize latent space distribution

d)

Minimize overfitting

55.

What differentiates Gradient Boosting from Bagging techniques like Random Forests?

a)

Use of decision trees

b)

Parallel training

c)

Sequential model training using residuals

d)

Sampling with replacement

56.

Which property does the softmax function NOT satisfy?

a)

Outputs are between 0 and 1

b)

Outputs sum to 1

c)

Preserves order of inputs

d)

Used in multi-class classification

57.

What is the role of the “critic” in Actor-Critic RL methods?

a)

Suggests next action

b)

Estimates value function

c)

Optimizes the environment

d)

Calculates policy gradient

58.

Which concept is central to the success of self-supervised learning?

a)

Data Augmentation

b)

Ensemble Learning

c)

Overparameterization

d)

One-hot Encoding

59.

Which of the following methods is used to interpret predictions from black-box models?

a)

Logistic Regression

b)

LIME

60.

Which AI approach focuses on mimicking the human brain structure and function?

a)

Symbolic AI

b)

Sub-symbolic AI

c)

Connectionism

d)

Cybernetics