Ph.D. Candidate Hao Tian Presents at ACM SIGSPATIAL 2026
CAI Lab Ph.D. Candidate Hao Tian attended the 34th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM SIGSPATIAL 2026), held November 3–6, 2026, in Riverside, California.
During the conference, Hao presented the lab's recent research, “BEACON: Behavioral and Semantic Enrichment of AlphaEarth Embeddings through Tri-Modal Contrastive Learning.”
The study introduces BEACON, a tri-modal contrastive learning framework that integrates Earth observation embeddings, POI information, and human mobility data to develop more human-centered urban representations. The framework improves predictions of socioeconomic and health-related outcomes while requiring only Earth observation embeddings during inference.
The conference provided a valuable opportunity to share our research, exchange ideas with researchers and practitioners, and explore emerging developments in GeoAI, geospatial foundation models, and urban computing.
We sincerely thank everyone who visited our poster and contributed to the engaging discussions. We look forward to future collaborations and continued exchanges with the SIGSPATIAL community!





