Matches in SemOpenAlex for { <https://semopenalex.org/work/W2010695026> ?p ?o ?g. }
Showing items 1 to 49 of
49
with 100 items per page.
- W2010695026 abstract "In the context of neural network procedures, it is proved that gradient descent on a surface defined by a sum of squared errors can fail to separate families of vectors. Each output is assumed to be a differentiable monotone transformation (typically the logistic) of a linear combination of inputs. Several examples are given of two families of vectors for which a linear combination exists that will serve to separate the two families. However, the minimum cost solution does not yield the desired combination. The examples include several cases where there are no local minima, as well as a one-layer system showing local minima with a large basin of attraction. In contrast to the perceptron convergence theorem, which proves that the perceptron architecture, there is no convergence theorem for gradient descent which would allow correct classification. The theorem disproves the presumption made in recent years, that barring local minima, gradient descent will find the best set of weights for a given problem. >" @default.
- W2010695026 created "2016-06-24" @default.
- W2010695026 creator A5034867895 @default.
- W2010695026 creator A5046060800 @default.
- W2010695026 creator A5068773239 @default.
- W2010695026 date "1988-01-01" @default.
- W2010695026 modified "2023-09-25" @default.
- W2010695026 title "Gradient descent fails to separate" @default.
- W2010695026 cites W80769075 @default.
- W2010695026 doi "https://doi.org/10.1109/icnn.1988.23902" @default.
- W2010695026 hasPublicationYear "1988" @default.
- W2010695026 type Work @default.
- W2010695026 sameAs 2010695026 @default.
- W2010695026 citedByCount "13" @default.
- W2010695026 crossrefType "proceedings-article" @default.
- W2010695026 hasAuthorship W2010695026A5034867895 @default.
- W2010695026 hasAuthorship W2010695026A5046060800 @default.
- W2010695026 hasAuthorship W2010695026A5068773239 @default.
- W2010695026 hasConcept C121332964 @default.
- W2010695026 hasConcept C153258448 @default.
- W2010695026 hasConcept C153294291 @default.
- W2010695026 hasConcept C154945302 @default.
- W2010695026 hasConcept C2776637919 @default.
- W2010695026 hasConcept C41008148 @default.
- W2010695026 hasConcept C50644808 @default.
- W2010695026 hasConceptScore W2010695026C121332964 @default.
- W2010695026 hasConceptScore W2010695026C153258448 @default.
- W2010695026 hasConceptScore W2010695026C153294291 @default.
- W2010695026 hasConceptScore W2010695026C154945302 @default.
- W2010695026 hasConceptScore W2010695026C2776637919 @default.
- W2010695026 hasConceptScore W2010695026C41008148 @default.
- W2010695026 hasConceptScore W2010695026C50644808 @default.
- W2010695026 hasLocation W20106950261 @default.
- W2010695026 hasOpenAccess W2010695026 @default.
- W2010695026 hasPrimaryLocation W20106950261 @default.
- W2010695026 hasRelatedWork W1970131234 @default.
- W2010695026 hasRelatedWork W1998295157 @default.
- W2010695026 hasRelatedWork W2026383778 @default.
- W2010695026 hasRelatedWork W2053365402 @default.
- W2010695026 hasRelatedWork W2885390211 @default.
- W2010695026 hasRelatedWork W3035836947 @default.
- W2010695026 hasRelatedWork W3035991101 @default.
- W2010695026 hasRelatedWork W3101388328 @default.
- W2010695026 hasRelatedWork W4205393829 @default.
- W2010695026 hasRelatedWork W4287754448 @default.
- W2010695026 isParatext "false" @default.
- W2010695026 isRetracted "false" @default.
- W2010695026 magId "2010695026" @default.
- W2010695026 workType "article" @default.