Change to convergence criterion#149
Open
kaghi wants to merge 3 commits into
Open
Conversation
…construction error. Should work and converge around 8-10 iterations instead of oscillating until max iterations.
…ry and is minimized with a best-fit kernel, which the loop reaches EARLY before it overshoots tau_d. We therefore select the minimum-reconstruction-residual iteration to finalize (replacing the biexp fit residual, which selected a later kernel), and allow for the guard. This works by allowing the stop after RESIDUAL_PATIENCE consecutive iterations that dont change. Because the residual minimizes early and rises as tau_d overshoots, this stops shortly after the optimum. Overall, better kernel and same spike quality spike inference with fewer iterations! Also fixed formatting issues with the iteration-manager.ts
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Changed the reconstruction residual from a bi-exp error to a trace-reconstruction