Home / Learn / AI & Machine Learning

🤖 AI & Machine Learning

Learning from data. 51 topics, from beginner to advanced.

Artificial intelligence and machine learning let software improve from examples instead of hand-written rules. This area explains how models are trained, how they are evaluated for overfitting and bias, and how neural networks, language models and reinforcement learning work. It is a practical guide to the ideas behind modern AI systems.

Who this area is for

Developers who want to build machine learning features, students of data science, and professionals who need to evaluate AI claims.

Real-world use cases

Classifying emails as spam or not spam from labeled examples.
Forecasting demand from past sales and seasonal patterns.
Recommending products based on what similar users bought.
Predicting customer churn so that a retention team can contact the right users.
Detecting unusual card transactions that may be fraudulent.

Topics in AI & Machine Learning

Beginner (17)

What is Artificial Intelligence — Building systems that perform tasks requiring human-like intelligence.
Machine Learning Overview — Algorithms that improve from data instead of explicit rules.
Supervised Learning — Learning from labeled examples to predict outputs.
Unsupervised Learning — Finding structure in unlabeled data.
Reinforcement Learning Basics — Learning actions by maximizing reward from an environment.
Training and Test Sets — Separating data used to learn from data used to evaluate.
Features and Labels — Input attributes and the target to be predicted.
Linear Regression — Fitting a straight line to predict continuous values.
Logistic Regression — Linear model producing class probabilities.
Decision Trees — Models splitting data by questions into a tree of decisions.
k-Nearest Neighbors — Classifying by majority vote of the closest examples.
k-Means Clustering — Partitioning data into k groups around centroids.
Overfitting and Underfitting — Model too tailored to training data versus too simple to learn.
Bias-Variance Tradeoff — Balance between systematic error and sensitivity to data.
Evaluation Metrics — Accuracy, precision, recall and F1 for classifiers.
Confusion Matrix — Table of correct and incorrect predictions per class.
Cross-Validation — Rotating held-out folds to estimate generalization.

Intermediate (17)

Gradient Descent — Iteratively stepping against the gradient to minimize loss.
Loss Functions — Quantities measuring how wrong predictions are.
Regularization — Penalties discouraging overly complex models.
Naive Bayes — Probabilistic classifier assuming feature independence.
Support Vector Machines — Classifiers maximizing the margin between classes.
Random Forests — Ensembles of decision trees voting together.
Gradient Boosting — Sequentially adding models that correct earlier errors.
Feature Engineering — Crafting inputs that make learning easier.
Dimensionality Reduction — Compressing features while preserving structure.
PCA — Projecting data onto directions of maximal variance.
Hyperparameter Tuning — Searching settings that control training.
Neural Networks — Layered networks of weighted units learning representations.
Perceptron — Simplest single-unit linear classifier.
Activation Functions — Nonlinearities such as ReLU and sigmoid.
Backpropagation — Computing gradients layer by layer via the chain rule.
Convolutional Neural Networks — Networks using shared filters for images.
Recurrent Neural Networks — Networks with loops handling sequences.

Advanced (17)

LSTM — Recurrent cells with gates preserving long-term information.
Attention Mechanism — Weighting which parts of input matter for each output.
Transformers — Attention-based architecture underpinning modern language models.
Large Language Models — Huge transformers trained to predict and generate text.
Embeddings — Dense vectors capturing meaning of items.
Transfer Learning — Reusing a pretrained model for a new task.
Generative Adversarial Networks — Generator and discriminator trained in competition.
Diffusion Models — Generative models learning to denoise random noise into data.
Autoencoders — Networks compressing then reconstructing their input.
Natural Language Processing — Computational handling of human language.
Tokenization — Splitting text into units for models.
Search Algorithms in AI — Exploring state spaces for goal states.
Minimax and Alpha-Beta — Game-tree search with pruning.
Bayesian Networks — Graphical models of probabilistic dependencies.
Reinforcement Learning Algorithms — Q-learning and policy gradients.
AI Ethics and Fairness — Addressing bias, transparency and societal impact.
MLOps — Practices for deploying and monitoring ML in production.

How to study AI & Machine Learning

Train one small model on a dataset you understand, such as house prices or flower measurements. Split the data, measure the error on unseen examples, and deliberately try to overfit it. Then add regularization and compare the results.

Frequently asked questions

Do I need advanced math to start machine learning?

You need basic algebra, probability and some comfort with vectors. Calculus and linear algebra deepen understanding but are not required to build and test your first models.

What is overfitting?

Overfitting happens when a model memorizes its training data, including noise, and performs poorly on new data. Held-out validation data and regularization are the standard defenses.

Is machine learning the same as AI?

Machine learning is one approach within AI. AI also includes rule-based systems, search and planning. Most current AI products rely on machine learning.

Can I learn AI without a powerful computer?

Yes. Small models and public datasets run on a laptop. Cloud notebooks are an option when a model needs more memory or a graphics processor.

How do I know if a model is good?

Measure it on data it did not see during training, using a metric that fits the task, such as accuracy, precision or recall. Compare it with a simple baseline before deciding it is useful.

Open the interactive AI & Machine Learning area Take the AI & Machine Learning quiz