Series

Machine Learning Foundations

A practical series for interviews, applied ML work, and AI/NLP research: data splitting, overfitting, cross-validation, metrics, model selection, hyperparameter search, leakage, and representations.

Start with data splits
中文
ML posts

Part I

Data Splits & Generalization

  1. 1. Train / Validation / Test Splits
  2. 2. Overfitting and Regularization
  3. 3. Cross-Validation for Model Evaluation
  4. 4. Bias–Variance Trade-off in Machine Learning

Part II

Evaluation & Selection

  1. 1. Metrics Beyond Accuracy
  2. 2. Model Selection: A Practical Guide
  3. 3. Grid Search and Randomized Search
  4. 4. Pipelines and Data Leakage

Part III

From Classical ML to Representation Learning

  1. 1. Feature Engineering vs Representation Learning

Source strategy

This series uses famous, reliable materials with clear roles instead of scattered blog posts.

Learn

Google ML Crash Course and ISLR/ISLP provide intuition, structure, and interview-friendly language.

Implement

scikit-learn gives practical APIs for splits, cross-validation, metrics, pipelines, and search.

Research

CS229, Deep Learning Book, and papers connect the basics to credible AI/NLP evaluation.

Build models without fooling yourself.

The goal is not just to know ML terms, but to make reliable training, evaluation, and selection decisions.

Read the first note
中文版本