AI Machine Learning Deep Learning Concepts Quiz

1. What is a recurrent neural network (RNN) designed to process?

Sequential data such as text or time series

Only static, independent data points

Image pixels without spatial relationships

Categorical variables exclusively

2. Which of the following is a key challenge in building AI systems?

Ensuring data quality, model interpretability, and fairness

Using as much data as possible regardless of relevance

Avoiding any model validation

Training only on historical data without updates

3. What is the purpose of 'normalization' in machine learning?

Scaling features to a standard range to improve model training

Removing all negative values from data

Converting categorical data to numerical

Reducing the number of training samples

4. A transformer is a deep learning architecture primarily used for ____.

Sequential data processing and language tasks

Image compression only

Hardware optimization

Database management

5. What does 'natural language processing' (NLP) focus on?

Enabling machines to understand and generate human language

Processing only numerical data

Compressing image files

Automating mechanical tasks

6. In the context of AI, what is 'bias'?

Systematic prejudice in data or algorithms leading to unfair predictions

A mathematical constant added to neural network weights

The total number of features used

The speed at which a model trains

7. What is 'transfer learning' in deep learning?

Using a pre-trained model on one task to improve performance on another

Moving data between different servers

Converting analog signals to digital format

Training multiple models on identical datasets

8. Which metric is most appropriate for evaluating a classification model?

Accuracy, precision, recall, or F1-score depending on the problem

Mean squared error only

The number of training epochs

The size of the training dataset

9. What is the primary goal of model validation?

Assess model performance on unseen data to prevent overfitting

Increase the number of training iterations

Simplify the model architecture

Eliminate all numerical features

10. In reinforcement learning, what is an 'agent'?

An entity that learns by interacting with its environment and receiving rewards

A labeled dataset used for training

A type of neural network layer

A person who supervises the learning process

11. What is artificial intelligence (AI)?

The ability of machines to perform tasks that typically require human intelligence

A type of computer hardware designed for gaming

Software that automates only simple, rule-based tasks

A programming language used exclusively for data analysis

12. A convolutional neural network (CNN) is typically used for which task?

Image recognition and processing spatial data

Predicting stock prices from time series

Translating text between languages

Clustering customer data by purchase history

13. What does 'backpropagation' accomplish in neural networks?

Computes gradients and updates weights to minimize error

Forwards data through the network for prediction

Removes irrelevant features from the input

Normalizes all input data to a standard scale

14. In a neural network, what is an activation function?

A mathematical function that introduces non-linearity to neurons

The process of training the network

The number of layers in the network

A method for storing training data

15. What is overfitting in machine learning?

A model memorizes training data too well and performs poorly on new data

A model fails to learn any meaningful patterns

Using too few features in the model

Training data that is too small for the algorithm

16. Which scenario best describes unsupervised learning?

Finding hidden patterns in unlabeled data without predefined outputs

Using labeled data to predict specific outcomes

Teaching a model to classify images with class labels

Requiring constant human feedback during training

17. What does 'supervised learning' require?

Labeled training data with known input-output pairs

No labels; the algorithm finds patterns independently

Human supervision during every prediction

Only numerical data with no categorical variables

18. In machine learning, what is a 'feature'?

An individual measurable property or characteristic used as input

The output prediction made by a model

A type of neural network activation function

The hardware used to run the algorithm

19. Deep learning is primarily based on what computational structure?

Artificial neural networks with multiple layers

Boolean logic gates

Linear regression models

Decision trees only

20. Which of the following best describes machine learning?

A subset of AI where systems learn patterns from data without explicit programming

The process of manually coding all decision rules into a computer

A method that only works with structured numerical data

An algorithm that requires human intervention for every prediction

SecAI+ Defining AI Machine Learning and Deep Learning Quiz