Publications
My research focuses on developing novel methods for spatio-temporal prediction with applications to urban systems, network science, and computational social systems. I am currently preparing manuscripts for submission to top-tier machine learning and AI conferences.
Preprints & Working Papers
Neural Time Capsule: Multi-Decadal Urban Growth Prediction via ConvLSTM-based Spatio-Temporal Deep Learning
Rohan Bali
In Preparation (Targeting AAAI 2026 or IJCAI 2026) (2025)
Urban growth forecasting is critical for infrastructure planning, resource allocation, and climate adaptation. We present Neural Time Capsule, a ConvLSTM-based deep learning architecture that predicts multi-decadal urban development patterns using historical satellite imagery. Our dual-channel approach processes both built-up surface density and road network infrastructure to capture spatio-temporal dependencies. Trained on 2,313 tiles spanning 25 years (1975-2000) across diverse U.S. regions, our model achieves 0.000218 MSE and demonstrates 67% improvement over U-Net baselines. The lightweight architecture (470K parameters) trains efficiently on CPU in 6 hours without requiring GPU resources, making it accessible for resource-constrained scenarios.