Research & Knowledge · AI and machine-learning development
AI categories, architectures and methods
Three separate registers. Each distinct architecture or method has one row; application domains and overarching concepts are kept apart. Historical project and paper initiatives remain linked without presenting proposals as completed research.
15 category and umbrella entries · 32 architecture and system-design entries · 14 method and algorithm entries. The original 40 base records are accounted for once each, with 6 additional research methods added.
Updated canonical taxonomy: The four-level register now separates learning paradigms from algorithms and architectures and tracks evidence status, aliases, and cross-references. Open the canonical four-level tables · Matching Excel-compatible CSV. The three legacy tables below remain for research continuity.
AI category types and domains
15 entriesApplication areas, capabilities and umbrella families. These are not counted as individual model architectures. Project and paper links identify representative proposed initiatives.
| # | AI type / method | Classification | Category or scope | Wikipedia | Manifold Grace project | Paper initiative | Research status |
|---|---|---|---|---|---|---|---|
| 1 | Computer vision | Application domain | Image and video understanding | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 2 | Natural language processing | Application domain | Language interpretation and generation | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 3 | Predictive analytics | Application domain | Prediction from historical and contextual data | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 4 | Expert and knowledge-based AI | AI approach / system domain | Symbolic knowledge and rule-guided reasoning | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 5 | Generative AI | Capability domain | Generation of text, images, audio and other outputs | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 6 | Multimodal AI | Capability domain | Joint learning or reasoning across modalities | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 7 | Agentic AI | System domain | Tool use, planning and multi-step actions | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 8 | Robotics and embodied AI | Application domain | Perception, motion and action in physical settings | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 9 | Graph and relational AI | Technical domain | Relationships, graph-structured data and inference | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 10 | Knowledge retrieval and question answering | Application domain | Source retrieval and answer generation | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 11 | AI-assisted decision and optimisation | Application domain | Ranking, inference and decision support | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 12 | Speech and audio AI | Application domain | Speech recognition, audio perception and synthesis | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 13 | Neural-network-based AI | Umbrella model family | A broad class containing many distinct architectures; not itself one unique architecture | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 14 | Classification and pattern recognition | Application domain | Predicting classes and identifying patterns | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
| 15 | AI-assisted process automation | Application / integration domain | Workflows that optionally incorporate AI models | Wikipedia | Project | Paper | Domain / umbrella; project link is a representative example, not an exclusive project |
Model and system architectures
32 entriesDistinct model classes and technical system designs already in the existing register. System integration patterns are labelled, not confused with neural model architecture.
| # | AI type / method | Classification | Category or scope | Wikipedia | Manifold Grace project | Paper initiative | Research status |
|---|---|---|---|---|---|---|---|
| 1 | CNN feature hierarchy | model architecture | Computer vision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 2 | Vision Transformer | model architecture | Computer vision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 3 | Object detection network | model architecture class | Computer vision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 4 | Semantic segmentation network | model architecture class | Computer vision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 5 | Instance segmentation network | model architecture class | Computer vision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 6 | Encoder Transformer | model architecture | Language and knowledge | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 7 | Autoregressive Transformer | model architecture | Language and knowledge | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 8 | Retrieval-augmented generation | system integration pattern | Language and knowledge | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 9 | Sequence-to-sequence Transformer | model architecture | Language and knowledge | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 10 | Knowledge-graph question answering | application architecture | Language and knowledge | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 11 | Temporal forecasting Transformer | model architecture | Predictive and tabular | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 12 | State-space sequence model | model architecture | Predictive and tabular | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 13 | Graph convolutional network | model architecture | Graphs and relational data | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 14 | Graph attention network | model architecture | Graphs and relational data | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 15 | Relational graph neural network | model architecture | Graphs and relational data | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 16 | Temporal graph neural network | model architecture | Graphs and relational data | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 17 | Neural relational database model | research model/system concept | Graphs and relational data | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 18 | Denoising diffusion model | generative model architecture | Generative and multimodal | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 19 | Latent diffusion model | generative model architecture | Generative and multimodal | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 20 | Generative adversarial network | generative model architecture | Generative and multimodal | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 21 | Variational autoencoder | generative model architecture | Generative and multimodal | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 22 | Vision-language model | multimodal model architecture class | Generative and multimodal | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 23 | Tool-using LLM agent | agent system architecture | Agents and automation | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 24 | Plan-and-execute agent | agent orchestration pattern | Agents and automation | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 25 | Multi-agent orchestration | multi-agent system pattern | Agents and automation | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 26 | AI-enabled BPM workflow | business workflow architecture | Agents and automation | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 27 | Event-driven decision agent | agent system architecture | Agents and automation | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 28 | Vision-based robotic policy | robotics policy architecture | Robotics and control | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 29 | Spiking neural controller | neural control architecture | Robotics and control | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 30 | Rule-based expert system | symbolic AI system | Logic, search and hybrid decision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 31 | Bayesian network | probabilistic graphical model | Logic, search and hybrid decision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 32 | Neuro-symbolic reasoning system | hybrid reasoning architecture | Logic, search and hybrid decision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
Statistical methods, learning paradigms and algorithms
14 entriesMethods and learning approaches, including survival analysis. A method may be used inside several architectures but is not itself an architecture just for that reason.
| # | AI type / method | Classification | Category or scope | Wikipedia | Manifold Grace project | Paper initiative | Research status |
|---|---|---|---|---|---|---|---|
| 1 | Gradient-boosted trees | machine-learning algorithm family | Predictive and tabular | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 2 | Generalised additive model | statistical model family | Predictive and tabular | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 3 | Survival and hazard model | statistical model family | Predictive and tabular | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 4 | Model-predictive controller | control method | Robotics and control | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 5 | Reinforcement learning policy | learning-policy method | Robotics and control | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 6 | Imitation learning policy | learning-policy method | Robotics and control | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 7 | Constraint-satisfaction solver | constraint-solving algorithm | Logic, search and hybrid decision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 8 | Case-based reasoning system | reasoning method | Logic, search and hybrid decision | Wikipedia | Project | Paper | Proposed project and paper; no independent validation claimed |
| 9 | Survival analysis | Statistical time-to-event method | Time-to-event and censored observations | Wikipedia | Project | Paper | Proposed research initiative; not a unique model architecture |
| 10 | Fuzzy logic | Logical reasoning framework | Approximate reasoning with degrees of truth | Wikipedia | Project | Paper | Proposed research initiative; not a unique model architecture |
| 11 | Supervised learning | Machine-learning training paradigm | Learning from labelled examples | Wikipedia | Project | Paper | Proposed research initiative; not a unique model architecture |
| 12 | Unsupervised learning | Machine-learning training paradigm | Finding patterns without target labels | Wikipedia | Project | Paper | Proposed research initiative; not a unique model architecture |
| 13 | Reinforcement learning | Machine-learning training paradigm | Sequential decisions with reward feedback | Wikipedia | Project | Paper | Proposed research initiative; not a unique model architecture |
| 14 | Deep learning | Machine-learning approach | Learning representations using multilayer models | Wikipedia | Project | Paper | Proposed research initiative; not a unique model architecture |
Research evidence and limits
The classification is an evidence-based working taxonomy, not a universal canonical count. A few system architectures can also be described as methods at a different level; the register assigns one primary home for counting and records that qualification. All proposed project and paper links remain working templates, not publications.
Survival analysis, censoring and Cox regression (2023) · Supervised algorithms · Unsupervised algorithms · NLP domain · AI/ML/deep learning distinctions.
Audit date: 14 November 2026. Original 1,000 proposed configurations preserved separately. Links to Wikipedia are introductory references, not independent proof of research validity.