← Back to catalogue
Research draft

inference

vr.tr.inference · ACT.PRC

Let an agent handle inference by type, premises or data, method, validity and uncertainty.

Thing Registry Activities and processes

Research draft, second pass

A second pass drafted this model: the structure a model of this thing needs, and what is known about it in the world. The line under this one says how the second half was obtained - researched against sources, or recalled without web access, in which case nothing here was read anywhere and every claim is a lead to verify. Unreviewed either way.

written by Claude from model knowledge without web access - no source was read, every claim is a lead to verify

Researched by: Claude

Purpose and description

Let an agent handle inference by type, premises or data, method, validity and uncertainty.

The process of deriving conclusions from premises or evidence, by deduction, induction or abduction in logic, and by statistical methods such as estimation and Bayesian inference in data analysis.

What it is for: Reasoning from what is known to what follows or is likely.

It can be check logical validity; estimate parameters from data; quantify uncertainty; avoid common fallacies.

Distinguishing features

From premises or data to conclusions

Deductive, inductive, abductive

Statistical inference quantifies uncertainty

Validity differs from truth

What it looks like

Arguments, proofs, statistical models and results.

How it is recognised

Premises and conclusions

Estimates with confidence or credible intervals

Machine learning inference means running a trained model

Related models

is a kind of - category

process and reasoning

is studied by - fields

logic and statistics

produces - outputs

conclusion and estimate

is confused with - another sense

model inference in ML

In practice

Families and kinds

deductive inference

inductive inference

abductive inference

frequentist statistical inference

Bayesian inference

Standards and regulation

Reporting guidelines for statistical analyses such as CONSORT

Failure modes and hazards

Fallacies

Overconfident estimates

p-hacking and multiple comparisons

Also called

Estimated Seismic Intensity Mapinferred from Commons categorypoint estimationBayesian inferenceBayesian post-hoc analysisPAC-bayesian learningBayesian inference in motor learningBayesian inference in phylogenyBayesian inference using Gibbs samplingexpectation propagationvariational Bayesian methodsknowledge inferenceinferred from page(s)inferred from URLinferred from DOI stringbackward chaininginferred from statementsinferred from NIOSH Numbered Publication ID matchinferred from proseimmediate inferenceinferred from referencesinferred from 'by descent' statementinferred from date official website first archivedinferred from date photo takeninferred from date of first (known) publicationinferred from date of last publicationinferred from system requirementsinferred from subtitleinferred from publication venueinferred from inclusion in a list or datasetinferred from place of family registerinferred from date of father's deathselective inferencenonparametric inferenceparametric inferenceinferred from date of mother's deathinformal inferential reasoningcut ruleecological inferenceabductive reasoning

+9

Where this came from

wikidata · CC0 1.0

Source repository Registry entry point

Drafted structure

Bundle to layer to finding to question, as the second pass will find it: 4 bundles · 8 layers · 8 findings · 16 questions.

Type Which inference.

Types differ.

Kind

Deductive to Bayesian.

Kind

Inference type.

  1. What kind of inference is used? definition
  2. What does it guarantee? boundary

Premises

Inputs.

Premises

Premises or data.

  1. What premises or data does it start from? provenance
  2. Are they reliable? boundary
Validity Soundness.

Validity must be checked.

Logic

Valid form.

Logic

Logical validity.

  1. Is the argument valid? boundary
  2. Does it commit a fallacy? boundary

Assumptions

Model assumptions.

Assumptions

Assumptions.

  1. What assumptions does the method make? definition
  2. Do they hold here? boundary
Uncertainty How sure.

Uncertainty must be stated.

Intervals

Confidence or credible.

Intervals

Intervals.

  1. What interval accompanies the estimate? measurement
  2. How is it interpreted? definition

Robustness

Sensitivity.

Robustness

Robustness.

  1. How sensitive is the conclusion to assumptions? measurement
  2. Were alternatives tested? provenance
Practice Reporting.

Good reporting prevents errors.

Reporting

Transparency.

Reporting

Reporting.

  1. Are methods and results reported fully? provenance
  2. Were analyses preregistered? provenance

Senses

ML inference.

Senses

Other senses.

  1. Is running a trained model meant instead? boundary
  2. Which entry fits? action

What the second pass must settle

  • Should inference types be separate entries?
  • How should the ML sense be split off?
  • How should uncertainty be recorded?
Draft as JSON