Core ML Concepts

This section covers the foundational machine learning theory, algorithms, metrics, and techniques that underpin every ML system — independent of any specific cloud platform or framework.

Topics

1. ML Problem Types — How to frame the question
2. Algorithm Selection — Choosing the right tool for the job
3. Neural Network Architectures — Deep learning building blocks
4. Data Preparation — Cleaning, scaling, and augmenting data
5. Feature Engineering — Encoding, dimensionality reduction, and multicollinearity
6. Training Concepts — Gradient descent, learning rate, loss functions, and distributed training
7. Hyperparameter Tuning — Strategies and key hyperparameters
8. Regularization — Preventing overfitting with L1, L2, dropout, and more
9. Evaluation Metrics — Classification, regression, and forecasting metrics
10. Overfitting vs Underfitting — Diagnosing and fixing model performance
11. Class Imbalance — Techniques for handling skewed datasets
12. Validation Strategies — Train/test splits, cross-validation, and time series splits
13. NLP Concepts — Text preprocessing and vectorization
14. Computer Vision — Image classification, object detection, and segmentation
15. Time Series — Components, rules, and forecasting approaches
16. Statistics and Probability — Distributions, correlation, and residual analysis
17. Model Explainability — SHAP, feature importance, and partial dependence