MétaCan
Menu
Back to cohort
Record W7042249772

Optimisation de boîtes noires multifidélités avec contraintes hiérarchisées

2023· other· fr· W7042249772 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.

fundA Canadian funder is recorded on the work.
aboutThe title or abstract carries a Canadian signal from the geographic lexicon.
no affNo Canadian affiliation: this work is invisible to an affiliation-only frame.
No Canadian affiliation. An affiliation-only frame, the usual design, would never have seen this work. It is one of the works that make the case for inverting the frame.

Bibliographic record

VenuePolyPublie (École Polytechnique de Montréal) · 2023
Typeother
Languagefr
FieldEngineering
TopicIoT-based Smart Home Systems
Canadian institutionsnot available
FundersNatural Sciences and Engineering Research Council of Canada
KeywordsLongest common subsequence problemNonlinear modelStatistical analysis
DOInot available

Abstract

fetched live from OpenAlex

RÉSUMÉ: On propose un algorithme d’optimisation de boîtes noires multifidélités qui s’intéresse au cas où une grande proportion du temps d’optimisation utilisé par les algorithmes de recherche directe est dépensé sur des points non réalisables. La méthode proposée doit être couplée avec un solveur existant, et elle permet à celui-ci de réduire le temps espéré des évaluations en estimant, à l’aide d’évaluations peu coûteuses, si un point est réalisable avant d’y investir plus de temps. Ces estimations sont obtenues avec une hiérarchisation des contraintes affectées par la multifidélité, qui est définie par une matrice de biadjacance. On propose une méthode de calcul de cette matrice. La recherche présentée s’inscrit dans le projet Alliance du Conseil de recherches en sciences naturelles et en génie du Canada (CRSNG) auquel participent Polytechnique Montréal et Hydro-Québec. À l’IREQ, un projet de recherche de stratégie de maintenance optimale requiert l’optimisation d’une boîte noire particulièrement coûteuse en temps qui contient plusieurs contraintes rarement satisfaites. L’algorithme est développé avec l’intention d’être appliqué à ce problème lorsque le projet d’Hydro-Québec atteindra cette phase. Dans le cadre de ce projet, des tests numériques sont effectués avec la famille de boîtes noires solar. Le solveur Optimisation non linéaire par recherche directe sur treillis adaptatifs : Nonlinear Optimisation by Mesh Adaptive Direct search (NOMAD) couplé à l’algorithme de hiérarchisation des contraintes est comparé au solveur NOMAD avec ses paramètres par défaut. Ces tests révèlent que l’algorithme permet de trouver des solutions significativement meilleures lorsqu’un point de départ réalisable est connu avant l’optimisation. Sans cette condition, les résultats sont variables; ils dépendent grandement des propriétés de la boîte noire optimisée. ABSTRACT: We propose a multi-fidelity blackbox optimization algorithm that addresses the problem of having to spend large computational resources on infeasible points when using direct search algorithms. The proposed method is coupled with an existing solver, allowing a decrease of the expected time per evaluation while keeping the efficiency and convergence properties of the existing method. This is achieved by estimating the feasibility of points to evaluate with low fidelity, hence low cost, evaluations before deciding if a point is worth the full time investment. These estimations are given by a hierarchy of constraints that are affected by the multi-fidelity, which is defined by a biadjacency matrix. We propose a computation method for this matrix. The project is part of the Natural Sciences and Engineering Research Council of Canada (NSERC) Alliance program. At Hydro-Quebec’s research institute : Institut de recherche en électricité du Québe (IREQ), an optimal maintenance strategy problem requires the optimization of a particularly costly blackbox, containing many rarely satisfied constraints. The proposed algorithm is developed with the intention of being applied to the optimal maintenance strategy problem when the project reaches the optimization stage. During this project, numerical tests are conducted on the solar family of blackbox problems. The Nonlinear Optimisation by Mesh Adaptive Direct search (NOMAD) software is used as the existing solver, and the solver with default parameters is compared to the solver coupled with the proposed algorithm during the tests. These tests show that given the same time budget, the coupling with the proposed method results in a great improvement in the quality of the solutions when a feasible starting point is known prior to the optimization. Without this condition, the results are mixed and largely depend on some properties of the optimized blackbox.

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)0.002
metaresearch head score (Gemma)0.001
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow), Research integrity, Insufficient payload (model declined to judge)
Consensus categoriesMeta-epidemiology (narrow), Insufficient payload (model declined to judge)
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: Simulation or modeling
GenreCandidate signal: Methods · Consensus signal: none
Teacher disagreement score0.608
Threshold uncertainty score1.000

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0020.001
Meta-epidemiology (narrow)0.0020.002
Meta-epidemiology (broad)0.0020.001
Bibliometrics0.0020.001
Science and technology studies0.0000.000
Scholarly communication0.0010.000
Open science0.0010.000
Research integrity0.0020.002
Insufficient payload (model declined to judge)0.0020.001

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 head0.010
GPT teacher head0.218
Teacher spread0.208 · 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