Learning-augmented Algorithms: Theory and Applications (LATA)

Talk

Infused Advice in Randomized Algorithms

Julien Dallot

at  15:30 ! Livein  Main Workshopfor  45min

Talk Abstract

An Infused Advice is a prediction directly injected into the algorithm’s random seed. This framework allows to study the effects of predictions on randomized algorithms in clean setting, without the need for error function or any other assumption on the prediction nature. I will present the model, and introduce analysis for a few classic problems.

 Overview  Program