Flow Optimization under Probabilistic Demands: Theory and Practice
Michael Schapira ( Hebrew University of Jerusalem )
- 14:00 9th November 2023 ( week 5, Michaelmas Term 2023 )Room 051
Flow optimization in the presence of uncertainty about future traffic demands is common in practice. A standard approach
to addressing this is predicting future traffic demands and optimizing with respect to these. This, however, can give rise to
poor quality solutions and prohibitively expensive runtimes. We present an alternative approach to this fundamental
challenge: direct stochastic optimization. We show, through theoretical analyses and extensive empirical evaluation,
that our approach yields both superior quality solutions and significantly faster runtimes.
Based on joint work with Yarin Perry, Felipe Vieira Frujeri, Chaim Hoch, Srikanth Kandula, Ishai Menache, and Aviv Tamar.
Awarded Best Paper at NSDI 2023
Share this:
