Areas
The site puts each page in one of 52 areas. An area has sub-areas. A grey tile has no pages yet.
- AI foundationsWhat AI is, what an agent is, and how we measure intelligence.
- Learning theoryWhen and why a learner can generalize from a finite sample.
- Math for AIThe mathematics that AI uses. Most items are bridge cards to math2.
- Optimization for learningHow a learner finds the parameters that minimize a loss.
- Search and evolutionary computationSearch through a space of states or candidates, and methods that copy natural evolution.
- Constraints and satisfiabilityProblems that a set of constraints defines, and the solvers that AI uses for them.
- Planning and schedulingHow an agent finds a sequence of actions that reaches a goal.
- Knowledge representationHow a system stores facts and rules, and how it reasons with them.
- Automated reasoningTheorem proving, logic programming and program synthesis in AI.
- Probabilistic reasoningReasoning under uncertainty with graphical models and inference.
- Classical AI systemsExpert systems, production rules, cognitive architectures and other systems before statistical learning.
- Games and multi-agent systemsGame playing, game theory and systems of many agents.
- Supervised learningLearning a map from inputs to labels with labelled examples.
- Unsupervised learningFinding structure in data with no labels.
- Probabilistic machine learningLearning with probability models, priors and uncertainty.
- Learning settingsSettings other than plain supervised learning, such as transfer, few-shot and continual learning.
- Time series and sequential dataForecasting, classification and anomaly detection on data in time order.
- Reinforcement learningHow an agent learns to act from rewards.
- Causal inference and discoveryCause and effect: identification, estimation and discovery of causal structure.
- Deep learning foundationsThe parts of a neural network and why a network can learn.
- Deep learning trainingThe methods that make the training of a deep network stable and efficient.
- Deep learning architecturesThe designs of neural networks, from feedforward networks to transformers and state space models.
- Generative modelsModels that learn a data distribution and sample from it.
- Self-supervised learningLearning representations from data with labels that the data itself supplies.
- Language modelsModels of text, from n-gram models to large language models.
- Post-trainingWhat changes a pretrained model into a useful and safe assistant.
- Reasoning and test-time computePrompting, chain of thought and other uses of compute at inference.
- Multimodal modelsModels that use more than one modality, such as text, images and audio.
- Efficient AIMethods that make models smaller, faster or cheaper.
- Foundation modelsThe families and releases of large pretrained models. This is the home of model nodes.
- Natural language processingThe tasks and methods of human language processing.
- Speech and audioSpeech recognition, speech synthesis, audio and music.
- Computer visionThe tasks and methods of image and video processing.
- Robotics and embodied AIAI in machines that sense and act in the physical world.
- AI agentsSystems that use models in a loop with tools, memory and plans.
- Retrieval and recommendationSearch, retrieval-augmented generation and recommender systems.
- AI for scienceAI in biology, chemistry, physics, climate, mathematics and other sciences.
- AI applicationsAI in medicine, code, law, finance, education, art and other fields.
- Data for AIHow training and evaluation data is collected, filtered, labelled and documented.
- ML systemsThe systems that train and serve models at scale.
- AI hardwareChips, accelerators and data centres for AI.
- AI softwareFrameworks, libraries, engines and tools built for AI.
- EvaluationMetrics, benchmarks and the methods that measure models. This is the home of metric and benchmark nodes.
- AI safety and alignmentHow to make AI systems do what their designers intend, and the risks when they do not.
- InterpretabilityMethods that explain what a model computes and why.
- Robustness, security and privacyAttacks on AI systems, the defences, and privacy in machine learning.
- Fairness and ethicsFairness, bias, transparency, accountability and other ethical questions of AI.
- AI governance and lawLaws, regulations, standards and policies about AI.
- History of AIThe history of AI from the early automata to the present.
- Philosophy and cognitive sciencePhilosophical arguments about machine minds, and the relation of AI to the study of the mind.
- AI and societyThe economics of AI and its effects on work, culture and public life.
- People and institutionsThe people, labs, companies, venues and awards of AI.