ML Fundamentals & Concepts
Core ML theory for SDE-ML, MLE, and AI Engineer interviews
Overview
This section covers the theoretical foundations of ML that top tech companies evaluate in concept-based interviews. You'll need to explain these concepts clearly and know when to apply them.
Document Structure
| Document | Focus |
|---|---|
| concepts-overview | How to approach ML concept questions |
| linear-regression | OLS, gradient descent, assumptions |
| logistic-regression | Binary classification, MLE, decision boundary |
| decision-trees | Splitting criteria, pruning, CART |
| svm | Margins, kernels, soft margin |
| knn | Distance metrics, k selection, curse of dimensionality |
| neural-networks | Backprop, activations, architectures |
| clustering | K-Means, DBSCAN, hierarchical |
| model-evaluation | Metrics, cross-validation, bias-variance |
Interview FAQ Section
Comprehensive Q&A guides with visualizations for the most commonly asked ML interview questions.
| Topic | Key Questions |
|---|---|
| Interview FAQ Index | Complete FAQ navigation |
| Bias-Variance Tradeoff | Error decomposition, learning curves |
| Regularization | L1 vs L2, dropout, early stopping |
| Gradient Descent | SGD, Adam, momentum, saddle points |
| Cross-Validation | K-fold, data leakage, stratified CV |
| Ensemble Methods | Bagging vs boosting, XGBoost |
| Hypothesis Testing | P-values, A/B testing, Type I/II errors |
Interview Question Types
| Type | What They Ask | Example |
|---|---|---|
| Algorithm Mechanics | How does X work? | "Explain how gradient descent works" |
| Trade-offs | When would you use X vs Y? | "Decision Tree vs Random Forest?" |
| Assumptions | What assumptions does X make? | "Linear regression assumptions?" |
| Failure Modes | When does X fail? | "When does KNN perform poorly?" |
| Hyperparameters | How do you tune X? | "How do you choose k in K-Means?" |
Study Order
Week 1: Supervised Learning Basics
- Linear Regression
- Logistic Regression
- Model Evaluation
Week 2: Tree-Based & Instance-Based 4. Decision Trees 5. KNN 6. SVM
Week 3: Deep Learning & Clustering 7. Neural Networks 8. Clustering (K-Means, DBSCAN)
Quick Reference: Algorithm Selection
| Problem Type | First Choice | When to Use |
|---|---|---|
| Regression | Linear Regression | Linear relationship, interpretability |
| Binary Classification | Logistic Regression | Interpretable probabilities |
| Multi-class | Random Forest | Non-linear, robust |
| High-dimensional | SVM with RBF | n_features > n_samples |
| Clustering | K-Means | Spherical clusters, known k |
| Density-based clustering | DBSCAN | Arbitrary shapes, outliers |