Explain Core ML Concepts
Company: Instacart
Role: Machine Learning Engineer
Category: Machine Learning
Difficulty: hard
Interview Round: Technical Screen
Answer the following machine learning interview questions:
1. Compare **linear regression** and **logistic regression**. Explain their goals, model outputs, loss functions, and typical use cases.
2. Explain the **bias-variance tradeoff** and how it relates to underfitting, overfitting, and generalization.
3. Suppose the loss function is **mean squared error (MSE)**. Will **full-batch gradient descent** and **stochastic gradient descent (SGD)** converge to the same point? Under what assumptions is the answer yes or no? Clarify the role of convexity, learning rate schedules, noise in SGD, and whether outliers make SGD inherently more robust than batch gradient descent when the objective is still MSE.
Quick Answer: This question evaluates understanding of core supervised learning concepts—comparing linear and logistic regression, interpreting the bias–variance tradeoff, and reasoning about optimization dynamics for MSE under full-batch and stochastic gradient descent—and is situated in the Machine Learning domain.