Skip to content

GCRI · Artificial Intelligence and Autonomous Systems

Statistics for AI evaluation

Probability and statistics make AI evaluation trustworthy: confidence intervals on scores, significance of differences between models, calibration of predicted probabilities and sampling designs for audits. In this field they stop small test sets and noisy metrics from being over-read.

01رجسٹر

Statistics for AI evaluation on the record.

Method recordSTEM method
  1. طریقہاحتمال اور شماریات
  2. Applied inArtificial Intelligence and Autonomous Systems
  3. Canonical pageSTEM Commons
02اطلاق

What probability and statistics does here.

Probability and statistics provide the mathematical framework for describing uncertainty and learning from data. They underpin estimation, testing and prediction in every risk discipline. In artificial intelligence and data and compute, statistical methods quantify the likelihood and uncertainty of losses across interconnected systems. They make the evidence behind risk decisions explicit and reproducible. The canonical description is on the STEM Commons.

Uncertainty on benchmark scores

Bootstrap and analytical intervals are reported with every score, and differences between models are tested before rankings are stated.

Calibration assessment

Predicted probabilities are compared with observed outcomes to check whether a model’s confidence matches its accuracy.

Audit sampling

Samples of decisions are drawn with stated designs so that error rates for subgroups can be estimated with known precision.

03Methods and evidence

How Statistics for AI evaluation is modelled, measured and tested.

The model, its assumptions, its data and its uncertainty, in the form a scientist or engineer would check. Structures follow published methods. The values drawn are illustrative, not findings.

Model, equations and assumptions

The model structure used for Statistics for AI evaluation, the assumptions it makes and how each is tested.

Risk of element iRi = f(Hi, Ei, Vi, Ci)hazard, exposure, vulnerability and coping capacity
Expected annual lossEAL = ∑k pk · Lkover events k with annual probability pk
Coupled systemsxt+1 = A xt + B ut + εtA carries the couplings, u the shock, ε the error
AssumptionWhy it is madeHow it is tested
Hazard and exposure are independent within an eventKeeps the loss integral tractableCompare with a copula fit on joint records
Couplings are linear near the operating pointLets shocks propagate through APerturb beyond the linear range and compare
Vulnerability curves transfer between placesLocal curves are rarely availableValidate against observed losses where they exist
Inputs
versioned datasets with provenance
Outputs
distributions, never single numbers
کوڈ
released with the model card and tests
Reproduction
same data, code, environment and seed
Standard risk and state-space formulations; the assumptions listed are typical of the method, stated for review, not results.
FAIR guiding principles · GO FAIR and FORCE11W3C PROV · World Wide Web ConsortiumGuidance on consistent treatment of uncertainties · IPCCCEOS Analysis Ready Data · Committee on Earth Observation Satellites
04ڈیٹا

What the method consumes.

Datasets and records used by probability and statistics on this platform, with the role each plays and the label its values carry. Sources are listed in the source register.

ڈیٹاکردارلیبل
Evaluation outputsScore estimationمشاہدہ شدہ
Labelled audit samplesError estimationمشاہدہ شدہ
Prediction logsCalibrationمشاہدہ شدہ
05عمل

عمل کیسے چلتا ہے۔

Process explorer · Process

Statistics for AI evaluation: Catastrophe model pipeline

10 steps5 stages

عمل

Statistics for AI evaluation: Catastrophe model pipeline

Catastrophe models usually have hazard, vulnerability, exposure and financial modules, and produce loss tables, average annual loss, exceedance curves, PML and TVaR.

  1. 01 Hazard

    1. 01Event catalogueDataset · Model developerThe hazard module holds a large catalogue of simulated events that represents a wide range of plausible scenarios.
    2. 03Hazard moduleModel · Model developerFor each event the hazard module gives the local hazard intensity that the vulnerability module reads.
  2. 02 Exposure and damage

    1. 02ExposureDataset · Incident responderThe exposure module houses the portfolio data, such as location-specific information.
    2. 04Vulnerability moduleModel · Model developerThe vulnerability module quantifies the expected damage to the exposures from an event.
  3. 03 Financial

    1. 05Financial moduleModel · Incident responderThe financial module calculates losses from all event scenarios. Ground-up, gross and net losses differ as shown below.
    2. 06Event loss tableDataset · Incident responderAn event loss table lists each event with its annual occurrence probability and its loss for the portfolio.
    3. 07Year loss tableDataset · Incident responderSome models also sample simulated years from event losses. This outline does not describe their format.
  4. 04 Metrics

    1. 08Exceedance probability curvesModel · Incident responderThe occurrence curve gives the chance that the largest loss in a year exceeds an amount. The aggregate curve does so for the sum of losses.
    2. 09AAL, PML and TVaRModel · Incident responderAverage annual loss sums event probability times loss. PML is one point on the occurrence curve.
  5. 05 Use of results

    1. 10Use of resultsDecision · Incident responderThe tail metrics are used when considering a portfolio’s exposure and the level of catastrophe reinsurance an insurer requires.

Outline of an external standard or method, from its public sources. It describes what the framework says, not a record of any run. It does not show that Nexus or any partner performs, certifies or approves a step. A gate is a step of the method, never a Nexus approval.

06توثیق

How results are checked.

Results are checked through diagnostics, cross-validation and sensitivity to assumptions. Independent replication confirms findings.

CheckHow it is done
Interval reportingEvery metric is reported with its uncertainty
Multiple comparisonsComparisons across many tasks are corrected for multiple testing
Sample size planningAudit samples are sized to detect differences of stated magnitude

Inference depends on assumptions about independence, distribution and sampling that may not hold. Small or biased samples limit reliability.

08نگہبانی

Who maintains the method.

Systems Science and Modelling stewards probability and statistics for the Consortium and reviews how it is applied across platforms.

09منسلک

یہ صفحہ پورے کنسورشیم میں کہاں لے جاتا ہے۔

10سوالات

سوالات و جوابات۔

What are the limits of probability and statistics in this field?
Inference depends on assumptions about independence, distribution and sampling that may not hold. Small or biased samples limit reliability.
How are results checked?
Results are checked through diagnostics, cross-validation and sensitivity to assumptions. Independent replication confirms findings.
Are the values on this page real?
The data names, roles and checks are real descriptions of practice. The canvas shows a design of the method’s use, not a run; no result is reported here.
Where else is this method applied?
The canonical method page lists every platform that applies it.
رپورٹیںحصہ ڈالیں

اس صفحے سے شواہد فراہم کریں۔

Reports کنسورشیم کی سہ ماہی اشاعت ہے۔ اس موضوع کے بارے میں آپ کا علم اگلے شمارے میں شواہد، کیس یا تصحیح بن سکتا ہے، جو آپ کے نام سے منسوب ہوگا اور اگر یہ آپ کا اپنا کام ہو تو CC BY 4.0 کے تحت شائع ہوگا۔

GCRIمصنوعی ذہانت اور خودکار نظامتیزی سے بڑھتی ٹیکنالوجیتجربہ گاہوں کے شواہد
مصنوعی ذہانت اور خودکار نظام

Bring a method, a dataset or a question.

معلومات اور استفسارات آج کھلے ہیں۔ استفسار کو ایک حوالہ نمبر ملتا ہے اور اگلے مرحلے کی نشاندہی ہوتی ہے؛ درخواستیں، ادائیگیاں، کمپیوٹنگ کی تخصیص اور رجسٹریشن صرف اس وقت کھلتی ہیں جب ان کی ٹیم، شرائط اور جائزے کی استعداد موجود ہو۔