Skip to content

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#software testing Open access Sep 2026

TraCE-SNN: Early Decisions and Compiler-Matched CPU Inference for Deletion-Aware Stroke Recognition

TraCE-SNN is a reproducible study of deletion-aware stroke recognition, early decisions and compiler-matched CPU inference. This version includes source code, frozen model checkpoints, predictions, experiment records, the manuscript and supplementary information. Version 1.1.0 adds an executable compiled prefix-replay path with in-loop confidence checks and immediate stopping. Reference agreement is verified for all prefix classes and early exits across five checkpoints. On the measured CPU, seed-11 mean request service time decreases from 50.32 to 41.48 microseconds (17.56%), with an empirical early-exit P99 of 73.69 microseconds. Raw request timings and the measurement protocol are included. Existing training runs and checkpoints are unchanged. The manuscript separates software service time from acquisition latency and reports the compiler-matched ANN comparisons, test-informed development history and external UJI evaluation. Raw datasets are acquired separately from their original sources. See REPRODUCING.md for commands and CHANGELOG.md for version changes.

Zichen Yu · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.