AI For Beginners 5 — Questions and Answers
Question 1: What is 'feature engineering' in a machine learning workflow?
- Selecting the best model architecture
- The process of creating or transforming input variables to improve model performance (Correct answer)
- Tuning hyperparameters using grid search
- Deploying a trained model to production
Correct answer: The process of creating or transforming input variables to improve model performance
Feature engineering transforms raw data into informative representations that help models learn patterns more effectively.
Question 2: Which algorithm is best described as building an ensemble of decision trees trained on random data subsets?
- Support Vector Machine
- Random Forest (Correct answer)
- Logistic Regression
- k-Means Clustering
Correct answer: Random Forest
Random Forest trains many decision trees on bootstrap samples and aggregates their predictions to reduce variance and improve accuracy.
Question 3: In AI model deployment, what is an API endpoint used for?
- Storing model weights on disk
- Exposing model inference capabilities as a network-accessible interface (Correct answer)
- Monitoring GPU memory usage
- Versioning the training dataset
Correct answer: Exposing model inference capabilities as a network-accessible interface
An API endpoint accepts input data over a network, runs it through the model, and returns predictions to the calling application.
Question 4: What is 'gradient descent' designed to minimize?
- Training time
- The number of model parameters
- A loss function measuring prediction error (Correct answer)
- The size of the dataset
Correct answer: A loss function measuring prediction error
Gradient descent iteratively moves model parameters in the direction that most steeply decreases the loss function.
Question 5: Which of the following best describes an AI 'pipeline'?
- A single neural network layer
- A sequence of data processing and modeling steps chained together end-to-end (Correct answer)
- A parallel computing hardware arrangement
- A database schema for storing predictions
Correct answer: A sequence of data processing and modeling steps chained together end-to-end
An AI pipeline connects steps like data ingestion, preprocessing, model training, evaluation, and deployment into a reproducible workflow.
Question 6: What is the primary purpose of normalization (e.g., min-max scaling) applied to input features?
- To increase the number of training samples
- To bring features onto a comparable scale so no single feature dominates training (Correct answer)
- To remove correlated features from the dataset
- To convert categorical variables into numbers
Correct answer: To bring features onto a comparable scale so no single feature dominates training
Normalization prevents features with large numerical ranges from disproportionately influencing gradient updates during training.
Question 7: What is 'model inference' in the context of a deployed AI system?
- The process of updating model weights with new data
- Running a trained model on new inputs to generate predictions (Correct answer)
- Evaluating a model on the validation set
- Selecting the best model from several candidates
Correct answer: Running a trained model on new inputs to generate predictions
Inference is the production-time step where a frozen, trained model processes new inputs and returns predictions without any weight updates.
AI For Beginners 5
What is 'feature engineering' in a machine learning workflow?