Machine Learning

About This Series

Machine learning is a technology that learns patterns from data to make predictions and decisions. This series begins with the foundations of statistical learning theory and progresses through classical methods, deep learning, and generative models. Emphasis is placed on balancing theoretical understanding—"why can machines learn?" and "when does learning succeed?"—with practical implementation and application.

The goal is not merely to use machine learning, but to become an engineer or researcher who can understand the principles and design systems accordingly. We study machine learning not as a black box, but as a technology grounded in mathematical foundations.

Learning by Level

Introductory

For those with programming experience

  • What is machine learning?
  • Supervised, unsupervised, and reinforcement learning
  • Intuitive understanding of learning
  • Setting up the Python environment

Basic

Undergraduate 1–2 level

  • Linear regression and least squares
  • Logistic regression
  • Decision trees and ensemble methods
  • Evaluation metrics and cross-validation

Intermediate

Undergraduate 3–4 level

  • Neural networks
  • Fundamentals of CNNs and RNNs
  • Optimization algorithms
  • Regularization and generalization

Advanced

Graduate level

  • Transformers and attention mechanisms
  • Generative models (VAE, GAN, Diffusion)
  • Statistical learning theory
  • Current research topics

Learning Path

Machine Learning Study Flow A diagram showing the learning path from Intro to Basic, Intermediate, and Advanced levels, with topics covered at each stage. Intro With programming exp. Basic Undergraduate 1-2 Intermediate Undergraduate 3-4 Advanced Graduate Intro: ML concepts, types of learning, Python setup Basic: Regression, classification, decision trees, evaluation Intermediate: Neural nets, CNN, RNN, optimization Advanced: Transformers, generative models, theory

Key Topics

Supervised Learning

Regression and classification. From linear models to nonlinear models and ensemble learning.

Unsupervised Learning

Clustering, dimensionality reduction, and anomaly detection. Discovering structure in data.

Deep Learning

Theory and practice of neural networks. CNNs, RNNs, and Transformers.

Generative Models

VAEs, GANs, and diffusion models. Learning to generate data.

Why Study the Theory?

Machine learning libraries make it possible to build something that "works." However:

  • Why can machines learn? What is the theoretical basis for generalization?
  • When does learning fail? Causes of overfitting and distribution shift.
  • How can we improve? Design rationale for regularization and data augmentation.
  • Handling new problems: Guidance when existing methods are insufficient.
  • Understanding current research: Essential for reading and implementing papers.

Answering these questions requires an understanding grounded in mathematical foundations.

Application Domains

  • Computer Vision: Image recognition, object detection, segmentation
  • Natural Language Processing: Machine translation, question answering, text generation
  • Speech Processing: Speech recognition, speech synthesis, speaker identification
  • Recommender Systems: Collaborative filtering, content-based methods
  • Scientific Research: Drug discovery, materials design, protein structure prediction
  • Autonomous Driving: Perception, prediction, planning

Related Series