BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:PIGNN-GPR: A Hybrid Machine Learning Framework for Spatio-Temporal
  PM2.5 Prediction - Reetha Thomas (Indian Institute of Management Bangalor
 e)
DTSTART:20250521T130000Z
DTEND:20250521T140000Z
UID:TALK232567@talks.cam.ac.uk
DESCRIPTION:Accurate prediction of pollutant concentration is essential fo
 r environmental protection and public health. Among various pollutants\, P
 M2.5 is particularly dangerous due to its severe health impacts\, making i
 ts precise forecasting a challenging task. In this work\, we propose a hyb
 rid Physics-Informed Graph Neural Network&ndash\;Gaussian Process Regressi
 on (PIGNN-GPR) framework for hourly PM2.5 prediction. The physics-informed
  layer incorporates the reaction-diffusion-advection equation to maintain 
 physical consistency\, while the Graph Neural Network captures spatial dep
 endencies using wind speed and direction across locations. To enhance pred
 iction reliability\, Gaussian Process Regression is used to refine outputs
  and provide uncertainty estimates. Additionally\, we apply Inverse Distan
 ce Weighting to interpolate PM2.5 levels at unmonitored sites. Model inter
 pretability is improved using SHapley Additive exPlanations\, identifying 
 the impact of key inputs like latitude\, longitude\, wind speed\, and dire
 ction. We validate our model with real-world data from the Delhi region. T
 he PIGNN-GPR framework offers both accurate and interpretable forecasts\, 
 aiding better air quality management.
LOCATION:Seminar Room 2\, Newton Institute
END:VEVENT
END:VCALENDAR