CSL SINE Group

View Full Calendar

CSL SINE SEMINAR - Victor Valls Delgado, Yale University

Event Type
Seminar/Symposium
Sponsor
Coordinated Science Lab
Location
141 CSL; 1308 W. Main Street, Urbana
Date
Jan 31, 2020   12:00 - 1:00 pm  
Speaker
Victor Valls Delgado, Postdoctoral Fellow, Yale University
Contact
Peggy Wells
Phone
217-244-2646
Views
73

Title: Online learning for network flow problems

Abstract: In this talk, I will present a convex optimization framework for solving online network flow problems: optimization problems where the rewards and constraints penalties are learned after the resource allocation decisions have been made. For instance, in data analytics problems, the utility obtained as a result of processing requests depends on the unknown information those contain, i.e., it is not known a priori. The framework is general and can readily encompass typical network features (e.g., time-varying network connectivity, time-varying flows, discrete actions, etc.) as well as different types of feedback on the rewards and constraints (i.e., deterministic, stochastic, and adversarial). In the second part of the talk, I’ll show how to use the framework to solve two specific problems: 1) resource allocation of data analytics with multi-grade service chains and bandit feedback (Infocom 2020); and 2) online ad placement with adversarial budget constraints (Aistats 2020).

Bio: I am a postdoctoral research fellow working with Prof. Tassiulas at Yale University. I obtained my degree from Universitat Pompeu Fabra (Barcelona) in 2011, and Ph.D. in applied mathematics from Trinity College Dublin in 2017, under the supervision of Prof. Leith. My research interests are in the area of mathematical optimization with applications in networks, control, and statistical learning. My research is supported by the European Union’s Horizon 2020 research and innovation programme under the Marie SkÅ‚odowska-Curie grant agreement No. 795244.

link for robots only