Instance-based Learning
Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary
A learning approach that builds no general model: it stores the training examples and classifies new inputs by comparing them to the closest stored cases.
What is Instance-based Learning?
Instance-based learning is a machine learning approach that builds no general model: it stores the training set and classifies each new input by comparing it to the most similar stored examples. Where other methods compress their data into parameters, an instance-based learner keeps the data and lets it speak directly.
It is also called lazy learning. Training costs almost nothing, because training is just storage; the work happens at prediction time, when the algorithm searches for neighbors.
How Instance-based Learning Works
Everything hinges on a similarity measure. A distance metric (Euclidean, Manhattan, or cosine) decides which stored examples count as close, and with the wrong metric or badly scaled features the neighbors are meaningless.
The main methods:
- k-Nearest Neighbors (k-NN): classifies a new instance by majority vote among its k closest stored examples; for regression, it averages their values.
- Locally weighted learning: weights nearby instances more heavily than distant ones, so predictions bend to local patterns.
- Case-based reasoning: retrieves whole past cases and adapts their solutions to the new situation, a style used in help-desk and legal-research systems.
The trade-off runs opposite to model-based learning. Adding knowledge is instant – store another example – but prediction slows as the dataset grows, and the stored data must stay available for as long as the system runs.
Example of Instance-based Learning
A wildlife app classifies animals from two measurements, weight and height, using k-NN with k set to 5. Its training set holds hundreds of labeled examples.
A hiker logs a new animal: 4 kg, 30 cm tall. The algorithm computes the distance from this point to every stored example and pulls the five closest: four labeled “mammal”, one labeled “bird”.
Majority vote says mammal, so that is the prediction. Nothing was learned in advance; the answer was assembled from raw stored cases at the moment of the query.
Related AI terms: K-Nearest Neighbors · K-means Clustering · Supervised Learning · Training Set
Did you like the Instance-based Learning gist?
Learn about 250+ need-to-know artificial intelligence terms in the AI Dictionary.
Mihail Sebastian — Writes about AI governance, regulation, and the technology behind them. Placeholder bio — replace with a real credential line. About