ML Anthology

Activation Atlas

Shan Carter, Zan Armstrong, Ludwig Schubert, Ian Johnson, Chris Olah
Distill 2019
doi:10.23915/distill.00015 /distill/2019/carter2019distill-activation/

Abstract

Distill articles are interactive publications and do not include traditional abstracts. This summary was written for the ML Anthology. Introduces activation atlases, a technique that combines dimensionality reduction with feature visualization to create explorable maps of what neural networks learn. Reveals how image classification networks organize visual concepts from simple textures to complex objects across layers.

Distill Code Semantic Scholar

Cite

Text

Carter et al. "Activation Atlas." Distill, 2019. doi:10.23915/distill.00015

Markdown

[Carter et al. "Activation Atlas." Distill, 2019.](https://mlanthology.org/distill/2019/carter2019distill-activation/) doi:10.23915/distill.00015

BibTeX

@article{carter2019distill-activation,
  title     = {{Activation Atlas}},
  author    = {Carter, Shan and Armstrong, Zan and Schubert, Ludwig and Johnson, Ian and Olah, Chris},
  journal   = {Distill},
  year      = {2019},
  doi       = {10.23915/distill.00015},
  url       = {https://mlanthology.org/distill/2019/carter2019distill-activation/}
}