explaining away

Tags: ml

• https://www.nature.com/articles/srep17531
• https://github.com/seansaito/Explaining-Away-Attacks-Against-Neural-Networks
• Explaining away is a pattern of reasoning that the confirmation of one cause of an obsereved event reduces the need to invoke other causes.

Example:

• A causes C

• B causes C

• C happened, A is observed

  • Implies B did not happen
  • However, B and A can happen at the same time.

• Sources

  • “The Notion of ‘explaining Away’ Evidence.” Accessed November 25, 2020. https://www.eecs.qmul.ac.uk/~norman/BBNs/The_notion_of__explaining_away__evidence.htm.
  • Cross Validated. “Probability - Why Does ‘Explaining Away’ Make Intuitive Sense?” Accessed November 25, 2020. https://stats.stackexchange.com/questions/54849/why-does-explaining-away-make-intuitive-sense.
  • Wellman, M. P., and M. Henrion. “Explaining ‘Explaining Away.’” IEEE Transactions on Pattern Analysis and Machine Intelligence 15, no. 3 (March 1993): 287–92. https://doi.org/10.1109/34.204911.
  • Moreno-Bote, Rubén, and Jan Drugowitsch. “Causal Inference and Explaining Away in a Spiking Network.” Scientific Reports 5, no. 1 (December 1, 2015): 17531. https://doi.org/10.1038/srep17531.