{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/4d514e2d6a0d40c78b4623be4333ee2b\" frameborder=\"0\" width=\"1152\" height=\"864\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":864,"width":1152,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":864,"thumbnail_width":1152,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/4d514e2d6a0d40c78b4623be4333ee2b-5d69f46bd721bb71.gif","duration":233.996,"title":"WaterwAIs","description":"This Loom explains how Riya Agrawal’s project Waterways uses AI to improve water infrastructure safety and efficiency. She focuses first on leak detection and anomaly detection, using unsupervised models like Isolation Forest, Autoencoder, and one-class SVM, each flagging 25 to 31 anomalies, and notes a supervised smart combined with XGBoost approach achieved over 75% overall accuracy but had zero recall on actual leak cases. For demand forecasting, she trained LSTM, Random Forest, profit, and an ARIMA plus machine learning hybrid, with Random Forest performing best at MAE around 4.33 and R square about 0.65. She also describes an early RL-based energy optimization module for pumping stations, using a simulated environment to reduce energy use while meeting demand."}