We are thrilled to announce that QREFM Lab members Duy Anh and Dat Le, under the supervision of Dr. Vuong Van Yen, have won the First Prize at the Singular Stochastic PDE Learning Competition. This event was co-organized by University College London (UCL) and the Academy of Mathematics and Systems Science.
This highly competitive event attracted over 70 teams, ranging from undergraduates to PhD researchers from world-leading universities, including Cambridge University, Imperial College London, and the University of Edinburgh.
The Challenge & Our Approach
Stochastic partial differential equations (SPDEs) driven by space–time white noise are fundamental mathematical models for physical phenomena such as turbulent flows and superconductivity. Recently, they have emerged as critical tools for modeling complex financial systems.
The QREFM Lab team proposed an innovative approach that integrates:
- The Neural SPDE framework [Salvi et al., 2022]
- Deep Latent Regularity Features [Gong et al., 2023]
This combination allowed the team to efficiently learn the mild solution of singular SPDEs. Furthermore, to address the autocorrelation issues commonly encountered in these solutions, the team designed a specialized Log-Spectral Loss in Fourier space.
Award
As the 1st-place team, QREFM Lab (NEU), alongside the 2nd-place team ("The Avengers", Cambridge University), will be invited to present their work and participate in a discussion session with the organizers. This opportunity opens doors for potential future research collaborations.
"For more detailed information about the competition and the leaderboard, please visit: Competition Details"