AI summaryⓘ
The authors created a system to help detect failures in optical networks more efficiently by learning from only a small amount of labeled data. Their method selectively asks for labels on only about 3.4% of incoming data, which saves resources while keeping accuracy very high. They also showed that this approach adds almost no delay compared to regular methods that don't adapt. This makes the system both fast and reliable for spotting problems as they happen.
active learningonline learningconcept driftoptical networksfailure detectionselective labelingaccuracyAUCstreaming datalatency
Authors
Yousuf Moiz Ali, Jaroslaw E. Prilepsky, João Pedro, Sasipim Srivallapanondh, Antonio Napoli, Sergei K. Turitsyn, Pedro Freire
Abstract
We propose a hybrid active-online learning framework for label-efficient concept drift adaptation in optical network failure detection. Using margin-based selective labeling, our method achieves nearceiling accuracy and AUC scores while querying only 3.4% of streaming samples, with negligible latency overhead compared to static inference.