Machine learning
Programs that improve at a task by finding patterns in data rather than being told every rule.
Also known as: ML · maschinelles Lernen
At a glance
Key signals
25% cross fields · reaches 27 more
- Machine Learning
- Artificial Intelligence
- Computer Science
- Explanation
- Examples
- Misconception
- Sourced relations 8
- Attribution
Dependencies
What this concept builds on and what it makes possible — derived from the atlas’s dependency, causal and structural relations, not from every related edge.
Foundations · builds on
Machine learningdepends onOptimizationEstablished
Learning is carried out as optimisation.
Mechanism: Training is posed as an optimisation problem: steadily change the model to reduce a measure of its error.
Machine learningdepends onPatternEstablished
Machine learning works by finding patterns.
Mechanism: A model generalises only if real patterns exist in the data; with no pattern, it can only memorise or guess.
Machine learningdepends onProbabilityEstablished
Machine learning reasons in probabilities.
Mechanism: Most models output likelihoods, not certainties, and are trained to make the observed data most probable.
Training dataenablesMachine learningEstablished
Training data is what a model learns from.
Mechanism: A model has nothing to learn until it is given examples; the training data is the raw material of learning.
Enables · leads to
Brain–computer interfacedepends onMachine learningEstablished
Open Brain–computer interface →
Decoding needs machine learning.
Mechanism: Models learn the mapping from noisy neural signals to intended movement or speech.
Machine learningenablesGeneralizationEstablished
Learning succeeds when the model generalises beyond its examples.
Mechanism: A model that only repeats its training examples has not learned; usable learning shows up as the capacity to generalise.
Protein folding predictiondepends onMachine learningEstablished
Open Protein folding prediction →
AlphaFold is deep learning.
Mechanism: The breakthrough came from a neural network trained on known protein structures.
System context
Machine learningis aAlgorithmEstablished
Machine learning is a family of algorithms.
Mechanism: A learning algorithm is still a precise procedure — but instead of fixed rules, it adjusts its own parameters from data.
Machine learningis part ofArtificial intelligenceEstablished
Open Artificial intelligence →
Machine learning is part of artificial intelligence.
Mechanism: Machine learning is a branch of AI in which systems improve at a task by learning patterns from data rather than following fixed rules.
- Wikidata verifiedmoderate evidence
Artificial neural networkis part ofMachine learningEstablished
Open Artificial neural network →
artificial neural network is part of Machine learning.
Mechanism: Neural networks are a family of machine-learning models: they learn by tuning the weights on their connections to reduce error.
- Wikidata verifiedmoderate evidence
Deep learningis part ofMachine learningEstablished
deep learning is part of Machine learning.
Mechanism: Deep learning is machine learning with many-layered neural networks, powerful enough to learn features directly from raw data.
- Wikipedia (English & German editions) verifiedmoderate evidence
Diffusion modelis aMachine learningEstablished
Generate by denoising.
Mechanism: A diffusion model learns to reverse noise step by step, generating images and audio.
Structural role & consequence
Interpreted from the current atlas graph — what the connections mean, not just how many there are.
Removing this node lengthens the structural route between Brain–computer interface and Protein folding prediction from 2 to 7 steps (they stay connected — alternative routes exist).
structural · Structural removal simulation — not a historical or causal counterfactual.
25% of its relationships cross field boundaries, reaching 27 other disciplines — unusual in a discipline where most concepts stay within their field.
structural · Structural graph analysis — not a claim of importance, causation or history.
Directly enables 1 concept; following enables/causes relations, 2 concepts are downstream across 5 disciplines.
structural · Follows only enables/causes dependency edges — not general relatedness.
Builds on 4 foundations (requires / depends-on / derived-from / emerges-from).
structural · Structural graph analysis — not a claim of importance, causation or history.
cross-field
15 within-field, 5 cross-field
Strengths & constraints
Constraints
- Evidence coverage currently thin in the atlas — few of its relationships carry claim-level evidence. atlas representation
- No dated history stored — the atlas records no key date for this concept. atlas representation
Conditions
- Its dependency reading rests on 4 foundation relations. structural
- Read structurally — most of its relationships carry no external evidence yet, so claims here are graph-derived. structural
Dependency radial
What this concept builds on (left) and what it makes possible (right) — derived from dependency and causal relations.
What builds on this
2 concepts build on this directly, 2 in total, across 4 disciplines.
Structural downstream reach along dependency edges — not a claim of historical necessity.
Seen through each discipline
How this concept sits in each of its fields — derived from its real connections in the graph, not asserted.
Through this lens it connects to Pattern, Artificial neural network, Supervised learning and Generalization.
Through this lens it connects to Artificial intelligence and Diffusion model.
Through this lens it connects to Algorithm, Artificial neural network, Supervised learning and Artificial intelligence.
Through this lens it connects to Probability, Generalization, Training data and VC dimension.
Where you'll meet it
Journeys that walk you through this idea. You may recognise it from more than one.
Related ideas to explore
Concepts that look related but are not yet connected here — candidates for a connection to reason about, not established links.
This idea also appears in…
The same structure shows up in other disciplines. These are real recurrences drawn from the graph — a starting point for asking “what carries over, and what changes?”
Information34 disciplines · 28 concepts
Probability23 disciplines · 21 concepts
Concepts
shares a mental model
shares a mental model
shares a mental model
shares a mental model
shares a mental model
shares a mental model
Instead of a programmer writing every rule, a machine-learning system is shown many examples and adjusts itself until it predicts well. It is pattern-finding at scale — powerful, but only as trustworthy as its data.
Mental models at work here
Information
Anything that reduces uncertainty. It can be encoded into a signal, sent across a channel, and decoded — and noise can corrupt it on the way.
Ask: what did the receiver not know before, and what does the message let them rule out?
Probability
A way to reason about uncertainty by assigning each possible outcome a share of the whole, between impossible (0) and certain (1).
Instead of 'will it happen?', ask 'how often would it happen if this repeated many times?'
- Connects 20 other ideas across 4 disciplines.
- A cross-disciplinary bridge — its connections reach into 27 other fields.
- Most of its connections are of the “Kind & structure” kind.
- It exercises 2 reusable thinking patterns.
- It features in 1 learning journey.
Derived from the graph’s real structure — observations, not a score.
Sources
- Machine learning (Encyclopaedia Britannica) verification pendingweak evidence
- Machine Learning, Statistics, and Data Analytics (2021) verified
- Statistics for machine learning (2012) verified