# Dictionary Learning **Domain:** Machine Learning / Signal Processing **Doc Type:** Technical Concept Node **Maturity:** Established **Related:** [[Sparse Autoencoder]], [[Monosemantic Feature]], [[Distributed Relational Compression]] --- ## Definition **Dictionary learning** discovers a set of reusable components whose combinations reconstruct observed signals or activations. Sparse coding favors explanations using relatively few dictionary elements for each observation. ## Architectural role The method provides one way to turn distributed internal geometry into inspectable features, enabling analysis or intervention without pretending the learned dictionary is the only valid decomposition.