Back to glossary

Deep learning

Derin öğrenme

D2

Deep learning is an approach to machine learning that uses models with layered representations of a target function.

Review status: 2027-08-28

Technical explanation

Deep feedforward networks learn a parameterized mapping from inputs to outputs through intermediate computations that form layers.

Conceptual boundaries

Deep learning is not synonymous with every neural network, because neural networks can have different structures and depth is a property of their layered composition.

Provider-neutral example

A team can use a deep feedforward model to turn image features into a suggested category while a person confirms the result.

Limitations

Approximating a target function is a modeling goal, not proof of accuracy in a specific deployment.

Related concepts

Atomic claims and evidence

  1. 1.1Goodfellow, Bengio, and Courville, Deep Learning, Introduction
    Source
    Goodfellow, Bengio, and Courville, Deep Learning, Introduction
    Source role
    Authoritative source
    Exact locator
    Introduction, p. 8: deep learning as an approach to AI and a type of machine learning
    Supported claim
    Deep learning is an approach to artificial intelligence and a type of machine learning, and deep feedforward networks are described as quintessential deep-learning models.
    Last verification
    Review due
    Scope limitation
    This identifies a major model class and does not make every AI system a deep-learning system.
    1.2Goodfellow, Bengio, and Courville, Deep Learning, Chapter 6
    Source
    Goodfellow, Bengio, and Courville, Deep Learning, Chapter 6
    Source role
    Authoritative source
    Exact locator
    Chapter 6, p. 164: Deep Feedforward Networks as quintessential deep-learning models
    Supported claim
    Deep learning is an approach to artificial intelligence and a type of machine learning, and deep feedforward networks are described as quintessential deep-learning models.
    Last verification
    Review due
    Scope limitation
    This identifies a major model class and does not make every AI system a deep-learning system.
  2. 2.1Goodfellow, Bengio, and Courville, Deep Learning, Chapter 6
    Source
    Goodfellow, Bengio, and Courville, Deep Learning, Chapter 6
    Source role
    Authoritative source
    Exact locator
    Chapter 6, p. 164: goal of a feedforward network and learned parameters
    Supported claim
    A deep feedforward network defines an input-to-output mapping and learns parameters that improve its approximation of a target function.
    Last verification
    Review due
    Scope limitation
    Approximation of a target function is a modeling goal, not proof of accuracy in every deployment context.