Swarm intelligence
Computational Paradigms AIMech:SwarmIntelligence DRAFT
A computational design paradigm, in the broadened 1999 formulation, that develops algorithms or distributed problem-solving devices inspired by collective behavior in social insects and other animal societies, emphasizing interactions among relatively simple agents from which system-level behavior emerges.
Identity Notes & Concept Scope
Selects the computational design paradigm in the inspected 32-page preview of Bonabeau, Dorigo and Theraulaz 1999. It organizes relatively simple agents and local interactions to obtain collective behavior. It is broader than an individual optimizer and does not require stigmergy in every implementation. The preview ends during the Introduction; complete-book coverage is not claimed. Existing owner 263 is reused. The fresh collection on 15 October 2026 at 16:12:44–16:13:52 UTC covers 769 native and 442 pinned AIO records at merged base 41221e5f8916ff024296a8ed34622bda702c276d, with only the five reviewed schema/renderer/docs/test edits applied. All 4,825 identity fields (3,687 normalized values), including local IDs and the explicit Sarsa lambda source spellings, were compared. No candidate identity collision was found. Four ignored- and hidden-inclusive checkout scans retained 4,432 occurrences in 367 exact contents. Root read all 304 newly encountered contents, all 11 contextual fields and nine neighboring definitions, and seven complete matching frontier owners. The supporting assessment read all 367 contents, including 63 historical ones, and reconstructed every occurrence. All 909 main and 393 nested frontier objects were mechanically queried. This is a dated, bounded identity assessment, not global absence or discovery saturation.
Claims & Grounded Evidence
5 Claims
Scope
#collective-design-scope
The 1999 formulation broadens swarm intelligence to the design of algorithms or distributed problem-solving devices inspired by the collective behavior of social insects and other animal societies. This selects an organizing approach rather than one prescribed computational procedure.
Mechanism
#individual-behavior-and-interaction
Designing the described collective behavior requires specifying both individual agent behavior and the interactions among agents or with their environment. The source’s self-organization account explains global patterns arising from interactions based on local information, rather than from a global pattern explicitly imposed on each constituent.
Scope
#engineering-modeling-distinction
The source distinguishes modeling a biological collective from designing an artificial problem-solving system. An engineering design may extend parameters beyond biologically relevant ranges or add nonbiological features while retaining useful organizing principles; biological plausibility is not the same criterion as engineering usefulness.
Mechanism
#direct-and-mediated-interactions
The inspected account permits direct interactions and indirect interactions in which an individual modifies the environment and later activity responds to that change. It identifies the latter pathway as stigmergy and discusses combining it with self-organization; the pathway is distinct from the broader swarm-inspired design approach.
Application
#algorithm-and-device-applications
The source situates the approach in combinatorial optimization, communication networks and robotics, and includes both algorithms and distributed problem-solving devices in its definition. These application classes do not supply a single shared learning objective, optimizer update or implementation protocol.
⚠️ Curated Limitations & Negative Results
- The selected identity follows the explicitly broadened 1999 formulation. The Beni and Wang 1993 cellular-robotics chapter was inspected only through its publisher abstract and metadata. Its full methods and the priority of earlier uses were not verified; the historical definitions are not asserted to be identical.
- The acquired source is a 32-page preview containing front matter and part of the introduction, ending midway through Section 1.4 at printed page 19. The complete introduction, later book chapters and the original experiments cited there were not inspected through this artifact; their detailed methods or quantitative results are not asserted.
- This organizing approach does not fix a complete executable algorithm. Agent state, interaction topology, scheduling, communication, stochastic rules, objectives and stopping conditions must be specified for a concrete design before its behavior or reproduction can be assessed.
- Flexibility, robustness, speed and other engineering benefits are motivations or conditional examples, not universal guarantees. The source emphasizes that programming emergent collective behavior is difficult; local rules alone do not establish convergence, optimality, fault tolerance or a favorable scaling law.
- Self-organization and stigmergy are distinct concepts, and the source permits direct as well as indirect interactions. Environmental traces, pheromone reinforcement, evaporation and reinforcement learning are not mandatory for every design within the selected scope; evidence about a particular application does not automatically generalize to the paradigm.
- The selected approach is not every multiagent system, distributed computation, biological collective phenomenon or population optimizer. A source-supported organizing relationship and an explicit boundary from particular implementations are needed; a shared name or metaphor alone does not establish membership or equivalence.
Artificial Intelligence Ontology (AIO) Grounding
Explicit AIO Snapshot Assessment: No Match
Rationale: None of these inspected classes captures the defining computational design approach or supplies a sufficiently informative necessary relation. Optional learning, model, network or mathematical implementations do not establish equivalence. This is a bounded assessment of the inspected pinned vocabulary, not all ontologies.
Search Scope: Current identity assessment covers 769 native and 442 AIO records at the merged cohort 344 base on 15 October 2026, with ignored and hidden files included in all four checkout searches. The pinned AIO commit is 397851c547d1c4d8f06ef184f63c1d40bc86937a. The retained source-specific mapping assessment examined the nine exact targets below and their 107 incident edges, alongside the separately documented native-neighbor context. Current identity collection and historical source/AIO reading retain distinct scopes. No informative mapping was established for this general concept in those inspected contexts; absence beyond this pinned vocabulary is not asserted.
| Examined AIO Candidate | Rejection Reason |
|---|---|
AIO:TrainingStrategy
|
The source approach includes distributed devices and robotics, not inherently preprocessing or training a machine-learning model. Training is a possible application, not the defining relation. |
AIO:MachineLearningTask
|
The inspected approach organizes computation and problem solving; learning from data is not required for every included system. |
AIO:ReinforcementLearning
|
Some routing or agent implementations learn from feedback, but no environment/reward-policy learning interface defines the whole swarm approach. |
AIO:UnsupervisedLearning
|
Grouping or sorting examples do not make the source-defined approach necessarily learn structure from unlabeled data. |
AIO:Model
|
The exact target is a generic abstract representation of properties and relationships. A model can inform a swarm design, but the design paradigm is not itself a particular representational model; that generic implementational link is not a useful identity mapping here. |
AIO:MathematicalFunction
|
The target is a general rule relating dependent and independent values, not limited to scalar arguments. A swarm design may contain functions, but no particular mathematical rule defines the broad approach; mere formalizability is insufficient. |
AIO:EvolutionaryLLM
|
The exact class is a large language model using evolutionary computation. Its related synonym evolutionary algorithms does not make every population or collective design an LLM. |
AIO:Network
|
The exact definition is broad: interconnected nodes or entities for communication, computation or exchange. Networks can implement swarm interactions, but this target does not capture the source-defined design principles; generic interconnection alone is insufficient for an informative local mapping. |
AIO:Layer
|
The exact class describes deep-learning topology between layers. The broad computational design approach is neither a neural layer nor its operation. |
Evidence Sources & Primary Literature
1 Source| Source ID | Publication / Document | Type / Access | Reference & Checksum |
|---|---|---|---|
bonabeau1999-preview
|
Eric Bonabeau, Marco Dorigo, Guy Theraulaz
|
primary monograph
full_text
|
ID: ISBN:9780195131581
sha256:d00cd88f5370…
|
Curation Provenance
Curator: Codex
Method: ai_assisted
Model: GPT-6
Created: 2026-10-15T16:28:32Z
Updated: 2026-10-15T16:28:32Z
Source File:
data/curated/SwarmIntelligence.yaml