We’ve moved the minimum confidence rating for classifiers / maximum anomaly score / maximum regression error etc. settings. Previously you could find these on the page for the learn block, but this was weird - these settings do not affect training the network, but rather affect how much we trust the learn block during live classification and model testing.
Thus, we’ve moved these settings to those pages. Go to either Live classification and load a sample, or go to Model testing, then click the three dots and select Set confidence thresholds. You’ll then be presented with a single overview from where you can change these settings for all learning blocks.
edit (25 July): We’ve reverted this build as it caused some test regressions, will re-land somewhere in the coming week!