Representation of features as images with neighborhood dependencies for compatibility with convolutional neural networks

Deep learning with Convolutional Neural Networks has shown great promise in image-based classification and enhancement but is often unsuitable for predictive modeling using features without spatial correlations. We present a feature representation approach termed REFINED (REpresentation of Features...

Descripción completa

Detalles Bibliográficos
Autores principales: Bazgir, Omid, Zhang, Ruibo, Dhruba, Saugato Rahman, Rahman, Raziur, Ghosh, Souparno, Pal, Ranadip
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Nature Publishing Group UK 2020
Materias:
Article
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7463019/
https://www.ncbi.nlm.nih.gov/pubmed/32873806
http://dx.doi.org/10.1038/s41467-020-18197-y

Internet

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7463019/
https://www.ncbi.nlm.nih.gov/pubmed/32873806
http://dx.doi.org/10.1038/s41467-020-18197-y

Ejemplares similares