Topics — What a work is about. Nucleation holds whatever reading the owner files here, whether or not an Anthology of the SOTA topic could also hold it; instruction for ML practice belongs to the anthology (ADR-048). The words are seeded from the clusters the reading-time triage of 2026-09-25 found, and grow by decision.
Mathematical statistics — the theory of statistical inference read for itself — comparison and sufficiency of experiments (Blackwell, Le Cam deficiency), statistical decision theory, the sufficiency theorems (Fisher–Neyman, Basu, Bahadur) and bounds on estimation. Bayesian modelling is probabilistic-modeling (ADR-031).
5 of 191 CASE documents. Back to the full index.
| # | Title | Summary | Status |
|---|---|---|---|
| CASE-042 v2 | Collaborative filtering and topic models: latent relational structure recovered from interaction matrices, up to gauge, or uniquely under anchor conditions | Proposed as a laboratory by the owner at U57 and conceded as prior art after the owner’s U58. Collaborative filtering recovers a user–item matrix’s predicted relations with latent coordinates fixed only up to an invertible change; topic models recover individual topics only under nonnegativity plus separability or anchor-word conditions. It shows that recovery is possible and limited, not that a recovered component is a communicative kind. | Active |
| CASE-087 | The discredited instrument and the second population: two pairs of readings with the same mean that the mean cannot tell apart once one instrument is discredited or a mixture is considered | A10 §2’s paired construction, which A10 recommended filing (“That single paired construction makes the dependence of sufficiency on the task and model very easy to see”). It grounds the relativity half of CLAIM-300. | Active |
| CASE-152 | Absent attempt or unresponsive receiver: with Y ∼ N(ab, 1), a the strength of an attempted influence and b the receiver’s responsiveness, the settings (a, b) = (0, 1) and (1, 0) give the same response distribution; K(a, b) = a²b²/2, the Fisher matrix has rank at most one and vanishes at the origin, and the real log canonical threshold is ½ with multiplicity 2, not the regular d/2 = 1, while changing b with a held fixed separates the two | The construction from the exchange of 2026-10-24 (workbench entry of 2026-10-24, U27), checked when filed. A two-parameter response model in which the observation depends on a product, the smallest singular model with a communicative reading. It shows a missing response failing to say whether nothing was attempted or nothing was taken up. | Active |
| CASE-159 | Authority geometry and response geometry: when an organization’s response is the authority-weighted mixture p_π(y | c) = Σ πᵢ pᵢ(y | c), the Fisher metric of its responses is the pullback JᵀD⁻¹J of the categorical (Shahshahani) metric on the shares π and never exceeds it; it vanishes along any redistribution between processes that respond alike under the probes used, so two specialists who both answer yes with probability 0.8 in routine contexts carry no information about their shares there, while a rare context in which they answer 0.9 and 0.1 carries 0.64/(p(1 − p)) per observation, p = 0.1 + 0.8π | From the continuation of the exchange of 2026-10-24 (U28–U31), where it appears in four replies (A28 §3, A29’s second priority, A30 §6A and A31’s probe diagnostic); checked when filed. The categorical metric on authority shares is the one in which the replicator is a gradient flow (CLAIM-444). What an evaluator sees is the response mixture, whose Fisher metric is that metric pulled back through the processes’ response profiles. Redistribution between processes that agree on the probes is invisible to them, and a probe on which they disagree restores it. The mixture counterpart of CASE-152’s product model. | Active |
| CASE-165 v4 | A dynamic beta–Bernoulli latent-feature model of an organization’s members: a finite Indian buffet prior over who holds which interpretive features, a sticky logistic transition for holding them, activations, and team, communication and authorization layers above; at the proposed size of sixteen features it falls well short of the Indian buffet process it quotes, its persistence term does not hold rare features, and the difference between having a capacity and using it is not identified from the outputs | The owner’s T22 (“maybe we could benefit from a restaurant process model, like an indian buffet model”), developed at T23 (“flesh out our proposed generating probability model further”) into a full generative specification for the Northstar firm, and carried to transformer training at T24. It is the candidate method of QUESTION-276 in its most worked-out form. Its facts are its definition, and four of them are not what T23 says: at K = 16 the prior expects 6.83 features, not αH₁₂ = 9.31; a held rare feature is kept with probability 0.16 a day; and having, using and splitting features are confounded in the reconstruction. It shows what such a model would have to get right before its features could count for anything. | Proposed |