English edition · 40 terms
Atlas of terms
Every English entry of the encyclopedia, by section: the mechanism behind each idea, an example on a well-known dataset and sources.
New to machine learning? Going step by step is easier: the ML course, 33 lessons.
By section
Foundations: maths and statistics
Supervised learning
- Decision tree
- Gradient boosting
- k-nearest neighbors (kNN)
- Linear regression
- Logistic regression
- Random forest
- What is machine learning
Model evaluation
- Bias–variance tradeoff
- Confusion matrix
- Cross-validation
- Overfitting and underfitting
- Precision and recall
- ROC AUC (area under the ROC curve)
Unsupervised learning
Neural networks
- Activation functions (ReLU, sigmoid, tanh)
- Backpropagation
- Dropout
- Learning rate
- Multilayer perceptron (MLP)
- Optimizers: SGD, momentum, Adam
- Perceptron (artificial neuron)
- Regularization (L2, dropout, early stopping)
Deep architectures
Large language models
- Large language model (LLM)
- Next-token prediction
- RAG — retrieval-augmented generation
- Temperature and sampling in LLMs
- Tokenization and BPE
- Word embeddings
Practice: from notebook to production
Laws and truths
A to Z
All 40 English terms from A to Z →
Reference datasets and learning paths
These pages are in Polish.