ml-deep-learning-hierarchical-features

IN premise — entries/2026/06/21/wiki-Machine_learning-chunk-3.md

Created 2026-06-21T09:55:50+00:00

Deep learning discovers hierarchical features where higher-level abstract features are built from lower-level ones across multiple layers

Dependents

These beliefs depend on this one:

• OUT deep-learning-fully-automates-representation — Deep learning fully automates representation learning through hierarchical feature discovery, eliminating the need for manual feature engineering in all practical settings.
• OUT deep-learning-hierarchy-sufficient-for-feature-learning — Deep learning's hierarchical feature discovery — bottom layers capturing raw patterns, intermediate layers building progressive abstractions, top layers composing task-relevant representations — would be sufficient to fully automate feature engineering across all domains.
• IN universal-approximation-vs-practical-depth — Universal approximation theory (Cybenko, 1989) establishes that shallow networks with sigmoidal activations can approximate any continuous function on compact subsets of R^n, but this existence guarantee has practical caveats: deep networks leverage hierarchical feature learning across multiple layers, and ReLU networks require width strictly greater than input dimension to maintain universal approximation — suggesting that theoretical approximation capacity alone does not determine architectural effectiveness.