Proceedings of 13th International Conference on Pattern Recognition 1996
DOI: 10.1109/icpr.1996.547656
|View full text |Cite
|
Sign up to set email alerts
|

Structural adaptation in mixture of experts

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
7
0
3

Year Published

1999
1999
2010
2010

Publication Types

Select...
7
1

Relationship

1
7

Authors

Journals

citations
Cited by 22 publications
(15 citation statements)
references
References 4 publications
0
7
0
3
Order By: Relevance
“…We have employed the localized gating network to solve a number of regression problems and have found that the network performs well with just a single layer of experts [20]. By requiring only one gating network, this model is significantly faster to train as compared to an hierarchical model.…”
Section: B a Localized Model For The Gating Networkmentioning
confidence: 97%
See 2 more Smart Citations
“…We have employed the localized gating network to solve a number of regression problems and have found that the network performs well with just a single layer of experts [20]. By requiring only one gating network, this model is significantly faster to train as compared to an hierarchical model.…”
Section: B a Localized Model For The Gating Networkmentioning
confidence: 97%
“…With such localized regions of expertise, a single layer of linear experts is adequate in practice for function approximation [20]. The gating network proposed in [14] is of the form (2) where (3) Thus the th expert's influence is predominant in a region around Note however that, because of the normalization performed in (2) the " "s are not strictly local, since they always need to sum to one.…”
Section: B a Localized Model For The Gating Networkmentioning
confidence: 99%
See 1 more Smart Citation
“…Moreover, as mentioned in Section III-C, model selection is a hard problem for the HME. Previous methods for model selection applied to the mixture of experts [20] and its hierarchical extension [22] has been addressed by several authors [35], [32], [30], [14]. However, all of these approaches are focused on the use of heuristic bottom-up procedures that start with a small network and grow it adaptively for learning the topology of the network.…”
Section: Discussionmentioning
confidence: 99%
“…Maiores detalhes concernentes a este algoritmo de otimização podem ser encontrados em [15] e em [5]. Assim, as equações do algoritmo wRLS adaptadas ao problema de otimização de são definidas como:…”
Section: Aprendizado Onlineunclassified