Methods for identifying emergent concepts in deep neural networks

The present perspective discusses methods to detect concepts in internal representations (hidden layers) of deep neural networks (DNNs), such as network dissection, feature visualization, and testing with concept activation vectors (TCAV). I argue that these methods provide evidence that DNNs are ab...

Descripción completa

Detalles Bibliográficos
Autor principal: Räz, Tim
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Elsevier 2023
Materias:
Perspective
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10318355/
https://www.ncbi.nlm.nih.gov/pubmed/37409048
http://dx.doi.org/10.1016/j.patter.2023.100761

Internet

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10318355/
https://www.ncbi.nlm.nih.gov/pubmed/37409048
http://dx.doi.org/10.1016/j.patter.2023.100761

Ejemplares similares