ECA value calculation

Hi all,

Can someone tell me how the programme calculate ECA values in 2D classification jobs?

Hi @aswathy! The ECA (Effective Classes Assigned) is the median ESS (Effective Sample Size) for particles in that class. A higher ESS means that the particle’s class assignment is less confident.

Formally, the ESS is the reciprocal of the sum of that particle’s square probabilities over all classes

\mathrm{ESS}(I_i) = 1 / (\sum_{k} p^2_{k(i)})

where:

  • I_i is the \mathrm{I^{th}} particle image
  • k is the class number
  • p_{k(i)} is the probability that I_i belongs to class k
1 Like

Great!Thank you so much.

1 Like