• The cost of searching for points in the neighborhood is O(N^2)
    • Humans may have hardware solutions.
  • Instead of making a similarity determination from all past data, we will only start with the most recent input.
    • That would make it O(N).
  • Even those that are not the most recent input are rarely successfully linked, but I’m not sure why that is.
  • Limiting memory to reasonable values will limit the amount of computation.

First published: 2019-12-17, edited slightly from 5da876a8aff09e0000cc6eb2.


This page is auto-translated from /nishio/記憶のサイズを制限する using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.