Research
My research focuses on:
Polynomial and semidefinite optimization
Exploitation of algebraic and convex structure to obtain stronger and more computationally efficient relaxations.
Christoffel–Darboux kernel methods
Strengthening low-order Moment–SOS relaxations, extracting candidate solutions, and guiding branch-and-bound algorithms.
Applications in machine learning and quantum optimization
Neural network verification, mixture model identification, and optimization problems arising from quantum systems.
Publications
Submitted
- Finite convergence of the Moment-SOS hierarchy under hidden convexity , 2026
- Leveraging the Christoffel–Darboux Kernel to Strengthen Moment–SOS Relaxations , 2025
Conferences
- Mixtures Closest to a Given Measure: A Semidefinite Programming Approach , accepted at ICML (oral), 2026 poster video slides
- Verifying Properties of Binary Neural Networks Using Sparse Polynomial Optimization , accepted at ICLR, 2025 poster
Journals
- On the Robustness Verification of Binary Neural Networks , Artificial Intelligence Review, 2026
Events
Conferences & Workshops
- ICML, Seoul, July 2026
- SIAM Conference on Optimization, Edinburgh, June 2026
- ICCV BiVision, Honolulu, October 2025
- ICCOPT, Los Angeles, July 2025
- ICLR, Singapore, April 2025
- LAAS Research and Technology Days, Toulouse, July 2024
- SinFra, Toulouse, June 2023
Summer/Winter Schools
- Moments, Non-Negative Polynomials, and Algebraic Statistics, University of Konstanz, Konstanz, February 2025
- TENORS Learning Week 1, Inria Université Côte d’Azur, Nice, February 2025
Teaching
- SC1015: Introduction to Data Science and Artificial Intelligence, 2024
- SC5010: Introduction to Data Analysis, 2023
Contact
Email: srecko.durasinovic@monash.edu or srecko001@e.ntu.edu.sg
Office: CNRS@CREATE Ltd., 1 CREATE Way, #08-01, CREATE Tower, Singapore 138602