MétaCan
Menu
Back to cohort
Record W2889264371 · doi:10.13033/isahp.y2018.049

SIGMOID SUPPLEMENTED DECISION STRUCTURES FOR EVIDENCE SENSITIVITY LEARNING

2018· article· en· W2889264371 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

VenueISAHP proceedings · 2018
Typearticle
Languageen
FieldComputer Science
TopicEducational Technology and Assessment
Canadian institutionsDefence Research and Development Canada
Fundersnot available
KeywordsSigmoid functionSensitivity (control systems)Computer scienceArtificial intelligenceArtificial neural networkEngineeringElectronic engineering

Abstract

fetched live from OpenAlex

During decision making for classification it is desirable to predict an outcome when not all of the evidence is available. Consider medical diagnosis. If a doctor is trying to determine the cause of a patient's ailment, often they are presented with a subset of potential evidences for or against a particular diagnosis. As the doctor runs more tests and the patients symptoms evolve, the doctor becomes more confident in their evaluation. It is critical that the decision maker be as confident in their decision as possible with as few evidences as are available. The goal of this paper is to improve the ability to predict the final decision given only a subset of the total information. By exploiting interdependencies and probabilistic relationships between the evidences, the confidence of prediction of a decision making tool can be improved through machine learning. Given some complete set of evidences, the Analytical Hierarchy Process (AHP) provides a method of weighting the nodes in a decision structure to synthesize a decision that reflects the opinion of a subject matter expert (SME). By truncating the comparison matrices produced for the AHP, weights can be generated for decision structures that are lacking inputs, known as deficient decision structures. This paper proposes a method of sigmoid node supplementation to the standard decision structure. Using machine learning the parameters of these sigmoid nodes can be optimized so that the output of deficient decision structures can be vastly improved for prediction of the output of the complete decision structure. This method preserves the original weights derived through the AHP and thus the relative importance of evidences is maintained after learning is undergone. An example will illustrate the improved confidence in prediction that can be achieved by adjusting the sensitivity of the supplemented sigmoid nodes.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.001
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Theoretical or conceptual · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.444
Threshold uncertainty score0.405

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.001
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0000.000
Scholarly communication0.0000.001
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.031
GPT teacher head0.346
Teacher spread0.315 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it