Incremental Learning Based on Growing Gaussian Mixture Models

Bouchachia, Abdelhamid; Vanaret, Charlie

doi:10.1109/icmla.2011.79

Cited by 27 publications

(28 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the context of the approach in [87], this is not considered as an undesired property, as the purpose for using these types of clusters is more to extract small micro-clusters, which, joined together, form one cluster of arbitrary shape. A similar consideration goes to the approach in [18], where Gaussian mixture models are grown for the same purpose (conducting micro-clustering).…”

Section: Methodsmentioning

confidence: 98%

“…Most of these extract ellipsoidal clusters in main position (axes-parallel). Some others (such as [29,18]) can model rotations in the data clouds, but are not equipped with fast merging operations. Thus, they are not able to resolve unnecessary overlaps or to dynamically compensate inappropriately chosen learning parameters (leading to over-clustering effects).…”

Section: Motivation and State-of-the-artmentioning

confidence: 99%

See 1 more Smart Citation

Autonomous data stream clustering implementing split-and-merge concepts – Towards a plug-and-play approach

Lughofer

Sayed‐Mouchaweh

2015

Information Sciences

View full text Add to dashboard Cite

Section: Methodsmentioning

confidence: 98%

Section: Motivation and State-of-the-artmentioning

confidence: 99%

Autonomous data stream clustering implementing split-and-merge concepts – Towards a plug-and-play approach

Lughofer

Sayed‐Mouchaweh

2015

Information Sciences

View full text Add to dashboard Cite

“…There are several possible approaches in the literature to deal with this problem, such as merge of similar components [6], [11], [12], the use of negative examples to eliminate concepts considered wrong and/or outdated [13] and removal information using criteria such as time or relevance.…”

Section: B Gaussian Component Removal Criteriamentioning

confidence: 99%

“…The incremental learning paradigm is a promising approach for learning in a data stream setting [1], [3]- [6]. In incremental learning, the classification model is updated for each new event, without the need to review all previous training examples.…”

Section: Introductionmentioning

confidence: 99%

IGMM-CD: A Gaussian Mixture Classification Algorithm for Data Streams with Concept Drifts

Oliveira¹,

Batista²

2015

2015 Brazilian Conference on Intelligent Systems (BRACIS)

View full text Add to dashboard Cite

Abstract-Learning concepts from data streams differs significantly from traditional batch learning, because in data streams the concepts to be learned may evolve over time. Incremental learning paradigm is a promising approach for learning in a data stream setting. However, in the presence of concept drifts, outdated concepts can cause misclassifications. Although several incremental Gaussian mixture models methods have been proposed in the literature, we notice that these algorithms lack an explicit policy to discard outdated concepts. In this paper, we propose a new incremental algorithm for data stream learning based on Gaussian Mixture Models. The proposed method is compared to various algorithms widely used in the literature, and the results show that it is competitive with them in various scenarios, overcoming them in some cases.

show abstract

“…O aprendizado incremental tem se mostrado uma abordagem promissora para o aprendizado em fluxo de dados (ZLIOBAITE, 2009;ENGEL;HEINEN, 2010;JÚNIOR, 2010;ELWELL;POLIKAR, 2011;BOUCHACHIA;VANARET, 2011). No aprendizado incremental, a hipótese de classificação é atualizada na presença de um novo evento, sem a necessidade de se rever todos os exemplos de treinamento anteriores.…”

Section: Introductionunclassified

Classificação de fluxos de dados não estacionários com algoritmos incrementais baseados no modelo de misturas gaussianas

Oliveira¹

View full text Add to dashboard Cite

Gostaria de agradecer à FAPESP (Fundação de Amparo à Pesquisa do Estado de São Paulo) pelo financiamento da pesquisa, à Universidade de São Paulo, por fornecer uma estrutura de excelência que com certeza fez com que este trabalho fosse desenvolvido da melhor forma possível e aos amigos que compartilharam experiências e conselhos. É muito importante também o agradecimento ao meu orientador, professor Gustavo E. A. P. A. Batista, pelo apoio durante toda a pesquisa, durante a qual ele esteve sempre disposto a ajudar, compartilhando seu conhecimento e apontando direções a serem seguidas, além de manter um caminho aberto para constante diálogo e troca de ideias.

show abstract

Incremental Learning Based on Growing Gaussian Mixture Models

Cited by 27 publications

References 10 publications

Autonomous data stream clustering implementing split-and-merge concepts – Towards a plug-and-play approach

Autonomous data stream clustering implementing split-and-merge concepts – Towards a plug-and-play approach

IGMM-CD: A Gaussian Mixture Classification Algorithm for Data Streams with Concept Drifts

Classificação de fluxos de dados não estacionários com algoritmos incrementais baseados no modelo de misturas gaussianas

Contact Info

Product

Resources

About