@HammerLabML‘s time to shine at the @WaterFuturesERC meeting @KIOSCoE. Barbara Hammer opens the floor with a quick update on research progress. Lots has already happened regarding #MachineLearning for decision making in water distribution systems, and much is yet to come.

Next, Paul Stahlhofen presenting his work on #AdversarialAttacks for water distribution networks. Bad news: models for critical infrastructure are vulnerable. 😱 Good news: Now we know, we can use this knowledge to make systems more robust. 💪

Inaam Ashraf shows how he uses #GraphConvolutionalNetworks to estimate water pressure, and detect and localize leaks. Results look promising, but have (still 😉) limited generalizability. Joint work with Luca Hermes, also from our group.

Last, Valerie Vaquet discusses #ConceptDrift. Sometimes, drift in the data distribution doesn’t trigger model drift, or the model drifts and produces false alarms. 🚨 Explanation approaches may shed light on drift in case of sensor faults and leakage localization.