Part V · 2 — Consolidated master matrices

draft

Single point of reference for the lookup matrices. The first three are those from Part I, doc 3 (there with the interpretive reading); the fourth synthesizes AI type × key sciences, derived from the 34 chapters of Part IV.

Intensity: ●●● central · ●● relevant · ● auxiliary · · rare Steps: 0 Problem · 1 Data · 2 EDA · 3 Modeling · 4 Training · 5 Evaluation · 5.5 Homologation · 6 Production · 7 Monitor. · 8 Retraining · 9 Governance


2.1 Mathematics × step

→ canonical table in ../01-life-cycle/03-sciences-and-mathematics.kmd §3.1. 17 sub-areas (linear algebra, calculus, optimization, probability, statistics, information theory, geometrytopology, graphs, groupssymmetries, combinatorics, logic, stochastic processes, measure, control, OR/queues, games, numbers/cryptography).

2.2 Other sciences × step

→ canonical table in §3.2. 15 sciences (physics, neuroscience, biology, cognitive, psychology, linguistics, economics, electrical eng., materials, control, philosophy, law, sociology, acoustics/optics, arts).

2.3 Technical inputs × step

→ canonical table in §3.3. 11 categories (CPU, GPUTPU, memorystorage, network, modeltrainingserving algorithms, data structures, systems/MLOps, cognitive concepts, modalities).


2.4 AI type × key sciences (synthesis of Part IV)

Each row summarizes the most salient sciences/mathematics of a type, per its chapter. (LAlinear algebra · Probprobability · Stat=statistics · Optoptimization · ITinformation theory · Ctrl=control)

Type (ch.) Dominant mathematics Dominant sciences
LLM (01) LA, Opt, Prob, IT Linguistics, neuroscience
Diffusion (02) Prob, stochastic processes, measure Physics (statistical mech.)
CNN (03) LA, groups/symmetries Optics/DSP, neuroscience
ViT (04) LA, Prob Linguistics+optics
GAN (05) Game theory, Prob, Opt
VAE (06) Prob (variational), IT
ASR (07) Prob, DSP Acoustics, linguistics
TTS (08) Prob, DSP Acoustics/psychoacoustics
Music (09) Prob, IT Music theory, acoustics
Video (10) Prob, geometry Physics/optics
GNN (11) Graphs, LA, symmetries Chemistry/biology
RNN (12) Calculus, time series
SSM (13) Ctrl, LA, DSP
Multimodal (14) LA, IT Linguistics+optics+acoustics
Expert System (15) Logic Knowledge engineering
Planning (16) Graphs, combinatorics, heuristics
Knowledge Graph (17) Description logic, graphs Linguistics/ontology
Solvers (18) Logic, complexity Philosophy (proof)
Bayesian Network (19) Prob, graphs, causality Statistics
HMM (20) Stochastic processes, Stat
Gaussian Process (21) Prob, LA, analysis Statistics
Classical ML (22) Stat, Opt, learning theory
Genetic Algorithm (23) Opt (gradient-free), Prob Evolutionary biology
Neuroevolution (24) Opt, Prob Evolutionary biology
Swarm (25) Opt, Prob Biology/ethology, complex systems
Deep RL value (26) MDP, dynamic programming, Prob Psychology (reinforcement)
AlphaZero (27) Games, search, MDP
RL Control (28) Ctrl, Opt, stochastic processes Physics
RLHF (29) Opt, decision/utility Psychometrics, ethics
Agent (30) Decision/games, logic Ctrl/cybernetics
RAG (31) LA, retrieval, IT Linguistics
Neuro-symbolic (32) Logic, combinatorics, Opt Philosophy/epistemology
Recommendation (33) LA, Stat, ranking Economics/games
Robotics (34) Ctrl, Prob, geometry Physics/mechanics

Visible pattern: linear algebra, probability and optimization cut across most types (the common base of ML); the specific sciences change with the modality (acoustics→audio, optics→vision, linguistics→text) and with the paradigm (logic→symbolic, biology→evolutionary, control→reinforcement/robotics).