Skip to main content
. 2026 Apr 29;28(5):502. doi: 10.3390/e28050502
Algorithm 3 Amortized Neural Inference.
  • Generate large number of synthetic QR datasets with known θ,γ

  • Compute summary statistics s for each dataset

  • Train NN fϕ(s)→(θ^,γ^) using MSE loss

  • For new data, compute summary statistics s∗ and predict θ^,γ^