AI/ML — The Complete Map

What Is AI/ML?

Machine Learning is a subset of AI where systems learn patterns from data instead of being explicitly programmed. Deep Learning is a subset of ML using neural networks with many layers.

AI (Artificial Intelligence)
 └── ML (Machine Learning)
      ├── Supervised Learning      labeled data → predict
      │    ├── Regression          predict a number
      │    └── Classification      predict a category
      ├── Unsupervised Learning    no labels → find structure
      │    ├── Clustering          group similar data
      │    └── Dimensionality Reduction   compress data
      ├── Deep Learning            neural networks with depth
      │    ├── CNN                 images
      │    ├── RNN / LSTM         sequences
      │    ├── Transformer        attention-based (LLMs)
      │    ├── GAN                generate data
      │    └── Diffusion          generate images
      └── Reinforcement Learning   learn by trial and error
           ├── Q-Learning         value-based
           ├── Policy Gradient    policy-based
           └── RLHF              align AI with humans

The Historical Arc

Every model exists because the previous one had a limitation. See Timeline for the full story.

Statistics (1800s)
  → Perceptron (1958) — "mimic the brain"
    → AI Winter — "it doesn't work"
      → Decision Trees, SVM (1980-90s) — "try something else"
        → Backprop revival (1986) — "neural nets can learn"
          → Deep Learning (2012) — "GPUs change everything"
            → Transformer (2017) — "attention is all you need"
              → LLMs (2020s) — "scale is all you need"

Learning Path

Start from what you use, drill down to why it works:

I use it — What does this model do in practice?
Why it exists — What problem did it solve that others couldn't?
Visualize — Interactive demo, see it work
Code — Implement it yourself
Math inside — The equation it optimizes
Math foundations — The prerequisite knowledge

Topics

Supervised Learning

Linear Regression — Where it all began
Polynomial / Ridge / Lasso — When a straight line isn't enough
Logistic Regression — Classification with probabilities
k-NN — Simplest classifier: follow your neighbors
Naive Bayes — Bayes' theorem in action
Decision Tree — Human-readable rules
Random Forest — Wisdom of crowds
Gradient Boosting / XGBoost — Kaggle's weapon of choice
SVM — The mathematically optimal boundary

Unsupervised Learning

k-Means — Find groups without labels
Hierarchical Clustering — Tree-shaped grouping
DBSCAN — Density-based clustering
GMM — Probabilistic clustering
PCA — Compress dimensions
t-SNE — Visualize high dimensions
UMAP — Faster, better t-SNE
Autoencoder — Neural compression

Deep Learning

Perceptron — The first neuron
MLP & Backpropagation — Learning to learn
CNN — See like a machine
RNN / LSTM / GRU — Remember sequences
Transformer — Attention is all you need
GAN — Create by competing
Diffusion Models — Create from noise
AlexNet → ResNet — The ImageNet revolution
BERT — Understanding language
GPT — Generating language
ViT — Transformer meets vision

Reinforcement Learning

Q-Learning — Learn by reward
DQN — Q-Learning + Neural Network
Policy Gradient / PPO — Learn the strategy directly
RLHF — Aligning AI with humans