Latent Dirichlet Allocation¶
A finite generative model gives each group Dirichlet-distributed mixture weights and each discrete item a latent component drawn from those weights.
Core Idea¶
Latent Dirichlet allocation is a finite probability model for grouped discrete items. Groups share component distributions; each group draws a Dirichlet mixture, and each item selects a component and then an observation from its distribution. The law exists before fitting or naming topics.[^ref-4735e085bd0c]
Scope of Application¶
Two documented habitats instantiate the same generative roles:
- Associated Press articles: articles are groups, word types the vocabulary, topic-word rows shared emissions, article proportions group mixtures, and token components item allocations; topic names remain interpretations.[^ref-4735e085bd0c]
- KTH action clips: clips are groups, quantized descriptor types the vocabulary, action-codeword rows shared emissions, clip proportions group mixtures, and descriptor components item allocations; codebook and component count are design choices. The study's 83.33% result belongs to pLSA, not LDA.[^ref-73e88aae7375]
Clarity¶
Shared emissions, group mixtures and item assignments are distinct; inference estimates them rather than defining the model.[^ref-4735e085bd0c] This LDA is neither supervised Linear Discriminant Analysis nor a Hierarchical Dirichlet Process without one fixed finite component count.[^ref-2722e4d2fc08]
Manages Complexity¶
Shared emissions let different groups reuse components while mixing them internally. The simple law is exchangeable: identical encoded counts have identical likelihood despite different order, so richer features or another law must preserve arrangement before fitting.[ref-4735e085bd0c][ref-73e88aae7375]
Abstract Reasoning¶
Name the groups, finite vocabulary, shared emissions, Dirichlet mixtures and item allocations, then check the conditional observation law.[^ref-4735e085bd0c] The whole is a Statistical Model, its approved parent, specialized by this grammar; neither a topic list nor a classifier supplies it.
Knowledge Transfer¶
Text-to-video transfer preserves probability roles, not word meanings. A visual codeword is a designed feature bucket; an inferred component can aid category selection without discovering a natural action class, so the supported reach remains grouped discrete modeling.[ref-4735e085bd0c][ref-73e88aae7375]
Relationships to Other Abstractions¶
Current abstraction Latent Dirichlet Allocation Domain-specific
Parents (1) — more general patterns this builds on
-
Latent Dirichlet Allocation is a kind of Statistical Model Domain-specific
A finite grouped Dirichlet allocation law is a statistical model with a shared emission family and per-item mixed-membership construction.
Hierarchy paths (4) — routes to 3 parentless roots
- Latent Dirichlet Allocation → Statistical Model → Representation → Abstraction
- Latent Dirichlet Allocation → Statistical Model → Probability Distribution → Random Variable → Function (Mapping)
- Latent Dirichlet Allocation → Statistical Model → Probability Distribution → Probability → Measure → Set and Membership
- Latent Dirichlet Allocation → Statistical Model → Probability Distribution → Random Variable → Probability → Measure → Set and Membership
Neighborhood in Abstraction Space¶
Latent Dirichlet Allocation sits in a moderately populated region (59th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.
Family — Codes, Matrices & Combinatorial Problems (32 abstractions)
Nearest neighbors
- Machine-Learning Model — 0.85
- Quadratic Assignment Problem — 0.85
- Microsegment — 0.85
- Factored Language Model — 0.84
- Covariance Matrix — 0.84
Computed from structural-signature embeddings · 2026-10-17
References¶
[^ref-4735e085bd0c]: David M. Blei, Andrew Y. Ng, and Michael I. Jordan, “Latent Dirichlet Allocation,” Journal of Machine Learning Research 3 (January 2003), 993–1022, full published paper. Definition and generative law: §3, printed pp. 996–997; AP illustration: §6, printed pp. 1007–1009; model comparisons and application limits: §§3.1, 5, 7–8. The published JMLR PDF was inspected. [^ref-2722e4d2fc08]: Yee Whye Teh, Michael I. Jordan, Matthew J. Beal, and David M. Blei, “Hierarchical Dirichlet Processes,” Journal of the American Statistical Association 101, no. 476 (December 2006), 1566–1581, doi:10.1198/016214506000000302, original author-hosted published paper. Abstract and §1, printed pp. 1566–1567 / PDF pp. 1–2: group Dirichlet processes share atoms from a discrete random base measure and do not predeclare one fixed finite component count. [^ref-73e88aae7375]: Juan Carlos Niebles, Hongcheng Wang, and Li Fei-Fei, “Unsupervised Learning of Human Action Categories Using Spatial-Temporal Words,” International Journal of Computer Vision 79 (2008), 299–318, doi:10.1007/s11263-007-0122-4, author-hosted journal-layout full text. Vocabulary: §3.2, PDF p. 7; LDA model and Eq. (7): §3.3.2, PDF pp. 8–9; KTH comparison: §4.1.1, PDF pp. 10–12. PDF page labels here are one-based and do not claim exact printed journal pagination.