LEAF: Navigating Concept Drift in Cellular Networks

Name: LEAF: Navigating Concept Drift in Cellular Networks
Uploaded: 2026-10-15T20:14:18.921548+00:00
Duration: 289 s
Description: LEAF is a new methodology for mitigating concept drift in cellular network models by using local error approximation. It detects drift, explains which features cause it, and then uses informed strategies like forgetting and over-sampling to retrain efficiently. This approach consistently outperforms standard retraining methods while significantly reducing the cost of retraining operations.

LEAF is a new methodology for mitigating concept drift in cellular network models by using local error approximation. It detects drift, explains which features cause it, and then uses informed strategies like forgetting and over-sampling to retrain efficiently. This approach consistently outperforms standard retraining methods while significantly reducing the cost of retraining operations.

Open the full episode