AAI Handbook
← All models
architecture

Multilayer Perceptron

intermediatesupervised
parametricclassificationregressionrepresentationtabular

An MLP stacks learned linear transformations and nonlinear activations to model complex feature interactions.

Mechanisms

dense layersbackpropagationnonlinear activation

Properties

nonlinearrepresentation learning

Constraints

requires large datarequires scalingsensitive to tuning

Practical profile

Explainability
low
Training cost
medium
Inference cost
low
Data appetite
high
Concepts to understand
BackpropagationBias and VarianceGeneralizationLoss and OptimizationThe PerceptronRegularizationTrain, Validation, and Test Sets