FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2149086872> ?p ?o ?g. }

• W2149086872 endingPage "20" @default.
• W2149086872 startingPage "1" @default.
• W2149086872 abstract "This article focuses on numerical optimization with continuous Estimation-of-Distribution Algorithms (EDAs). Specifically, the focus is on the use of one of the most common and best understood probability distributions: the normal distribution. We first give an overview of the existing research on this topic. We then point out a source of inefficiency in EDAs that make use of the normal distribution with maximum-likelihood (ML) estimates. Scaling the covariance matrix beyond its ML estimate does not remove this inefficiency. To remove the inefficiency, the orientation of the normal distribution must be changed. So far, only Evolution Strategies (ES) and particularly Covariance Matrix Adaptation ES (CMA-ES) are capable of achieving such re-orientation. In this article we provide a simple, but effective technique for achieving re-orientation while still only performing the well-known ML estimates. We call the new technique Anticipated Mean Shift (AMS). The resulting EDA, called Adapted Maximum-Likelihood Gaussian Model -- Iterated Density-Estimation Evolutionary Algorithm (AMaLGaM-IDEA) adapts not only the ML estimate for the covariance matrix, but also the ML estimate for the mean. AMaLGaM-IDEA has an improved performance compared to previous EDAs that use ML estimates as well as compared to previous EDAs that scale the variance adaptively. Also, we indicate the circumstances under which AMaLGaM-IDEA is found to be robust to rotations of the search space. A comparison with CMA-ES identifies the conditions under which AMaLGaM-IDEA is able to outperform CMA-ES and vice versa. We conclude that AMaLGaM-IDEA is currently among the most efficient real-valued continuous EDAs while at the same time it is relatively simple to understand (especially in the naive, univariate case). Pseudo-code is provided in this article; source-code can be downloaded from the web." @default.
• W2149086872 created "2016-06-24" @default.
• W2149086872 creator A5031539135 @default.
• W2149086872 creator A5032615368 @default.
• W2149086872 creator A5063460579 @default.
• W2149086872 date "2007-12-01" @default.
• W2149086872 modified "2023-09-23" @default.
• W2149086872 title "Adapted Maximum-Likelihood Gaussian Models for Numerical Optimization with Continuous EDAs" @default.
• W2149086872 cites W1507225143 @default.
• W2149086872 cites W1532567435 @default.
• W2149086872 cites W1534948419 @default.
• W2149086872 cites W1540706608 @default.
• W2149086872 cites W1541774237 @default.
• W2149086872 cites W1547363087 @default.
• W2149086872 cites W1548369215 @default.
• W2149086872 cites W1692958259 @default.
• W2149086872 cites W1966241132 @default.
• W2149086872 cites W1966497546 @default.
• W2149086872 cites W2005470683 @default.
• W2149086872 cites W2020009149 @default.
• W2149086872 cites W2058753054 @default.
• W2149086872 cites W2063698478 @default.
• W2149086872 cites W2101715222 @default.
• W2149086872 cites W2112036188 @default.
• W2149086872 cites W2129794603 @default.
• W2149086872 cites W2153459327 @default.
• W2149086872 cites W2156909104 @default.
• W2149086872 cites W2159072510 @default.
• W2149086872 cites W2162036626 @default.
• W2149086872 cites W2164383876 @default.
• W2149086872 cites W2167795535 @default.
• W2149086872 cites W2168764293 @default.
• W2149086872 cites W2278135314 @default.
• W2149086872 cites W2324309783 @default.
• W2149086872 cites W2801179766 @default.
• W2149086872 cites W82518839 @default.
• W2149086872 hasPublicationYear "2007" @default.
• W2149086872 type Work @default.
• W2149086872 sameAs 2149086872 @default.
• W2149086872 citedByCount "4" @default.
• W2149086872 countsByYear W21490868722012 @default.
• W2149086872 countsByYear W21490868722014 @default.
• W2149086872 crossrefType "journal-article" @default.
• W2149086872 hasAuthorship W2149086872A5031539135 @default.
• W2149086872 hasAuthorship W2149086872A5032615368 @default.
• W2149086872 hasAuthorship W2149086872A5063460579 @default.
• W2149086872 hasConcept C105795698 @default.
• W2149086872 hasConcept C11413529 @default.
• W2149086872 hasConcept C121332964 @default.
• W2149086872 hasConcept C126255220 @default.
• W2149086872 hasConcept C159149176 @default.
• W2149086872 hasConcept C162500139 @default.
• W2149086872 hasConcept C16345878 @default.
• W2149086872 hasConcept C163716315 @default.
• W2149086872 hasConcept C178650346 @default.
• W2149086872 hasConcept C185142706 @default.
• W2149086872 hasConcept C205555498 @default.
• W2149086872 hasConcept C207002847 @default.
• W2149086872 hasConcept C2524010 @default.
• W2149086872 hasConcept C33923547 @default.
• W2149086872 hasConcept C41008148 @default.
• W2149086872 hasConcept C49284225 @default.
• W2149086872 hasConcept C62520636 @default.
• W2149086872 hasConceptScore W2149086872C105795698 @default.
• W2149086872 hasConceptScore W2149086872C11413529 @default.
• W2149086872 hasConceptScore W2149086872C121332964 @default.
• W2149086872 hasConceptScore W2149086872C126255220 @default.
• W2149086872 hasConceptScore W2149086872C159149176 @default.
• W2149086872 hasConceptScore W2149086872C162500139 @default.
• W2149086872 hasConceptScore W2149086872C16345878 @default.
• W2149086872 hasConceptScore W2149086872C163716315 @default.
• W2149086872 hasConceptScore W2149086872C178650346 @default.
• W2149086872 hasConceptScore W2149086872C185142706 @default.
• W2149086872 hasConceptScore W2149086872C205555498 @default.
• W2149086872 hasConceptScore W2149086872C207002847 @default.
• W2149086872 hasConceptScore W2149086872C2524010 @default.
• W2149086872 hasConceptScore W2149086872C33923547 @default.
• W2149086872 hasConceptScore W2149086872C41008148 @default.
• W2149086872 hasConceptScore W2149086872C49284225 @default.
• W2149086872 hasConceptScore W2149086872C62520636 @default.
• W2149086872 hasLocation W21490868721 @default.
• W2149086872 hasOpenAccess W2149086872 @default.
• W2149086872 hasPrimaryLocation W21490868721 @default.
• W2149086872 hasRelatedWork W125706094 @default.
• W2149086872 hasRelatedWork W1492710783 @default.
• W2149086872 hasRelatedWork W1877740990 @default.
• W2149086872 hasRelatedWork W1966497546 @default.
• W2149086872 hasRelatedWork W2066331031 @default.
• W2149086872 hasRelatedWork W2129794603 @default.
• W2149086872 hasRelatedWork W2281047127 @default.
• W2149086872 hasRelatedWork W2559375353 @default.
• W2149086872 hasRelatedWork W2612998929 @default.
• W2149086872 hasRelatedWork W2743389983 @default.
• W2149086872 hasRelatedWork W2781321050 @default.
• W2149086872 hasRelatedWork W2792356241 @default.
• W2149086872 hasRelatedWork W2798756616 @default.
• W2149086872 hasRelatedWork W2951226321 @default.
• W2149086872 hasRelatedWork W2953059960 @default.

• next