Description
Course project for CS 274E: Deep Generative Models at UC Irvine (2025). It compares generative approaches to predicting the next day’s active fire mask from a history of fire observations: a conditional VAE for fast, stochastic forecasts with uncertainty, symbolic regression for interpretable equations of fire spread, and diffusion models for fine-grained spread synthesis, on a subset of the WildfireSpreadTS benchmark.