How to reach the venue?

The workshop is virtual at this link, but only for registered IJCAI participants. You can still register.

To get the Zoom link of the talks, you need to 'walk' to the Red Wing, South 1 room (the left-most room on the left-most wing in the left-most side of the map). Or type the name of one of the organizers, press 'locate' and follow the line on the floor.

Detailed program

All talks are 15m presentation + 5m Q&A.

To access the Zoom link, visit the IJCAI Gather.Town website and 'walk' to Red Wing, South 1. Press X to see the link.

Coffee breaks are in the Gather.Town space, where you can talk in small 'groups' like in a real workshop.

7 January (CET time)

9:00-9:10 Welcome to the virtual room and workshop intro

Session 1: Interplay between ML and optimization methods

9:10-9:30 Black-box Mixed-Variable Optimisation using a Surrogate Model that Satisfies Integer Constraints -- PDF

Laurens Bliek, Sicco Verwer and Mathijs De Weerdt

9:30-09:50 Machine learning for improving performance in an evolutionary algorithm for minimum path with uncertain costs given by massively simulated scenarios -- PDF

Ricardo Di Pasquale and Javier Marenco

09:50-10:10 Algorithm Selection across Selection Hyper-heuristics

Mustafa Misir -- PDF

break with informal chats: 10:10-10:40 in Gather.Town

Session 2: ML for solving optimization problem

10:40-11:00 Machine learning-aided optimization of baseband function placement for 5G Radio Access Networks

Ligia Maria Moreira Zorello, Laurens Bliek, Sebastian Troia, Guido Maier and Sicco Verwer

11:00-11:20 Understanding Reinforcement Learning Based Traveling Salesman Problem Solvers -- PDF

Tianyi Chen

Session 3: Optimization techniques for ML

11:20-11:40 On Training Neural Networks with Mixed Integer Programming -- PDF

Tómas Þorbjarnarson and Neil Yorke-Smith

11:40-12:00 Optimizing Hyperparameters in CNNs using Bilevel Programming for Time Series Data -- PDF

Taniya Seth and Pranab K. Muhuri

discussion session: 12:00-12:30 in Gather.Town

8 January (CET time)

9:00-10:00 Invited talk by Dolores Romero Morales from Copenhagen Business School:

Mathematical optimization in classification and regression trees

break with informal chats: 10:00-10:30 in Gather.Town

Session 4: Optimization techniques for ML

10:30-10:50 Optimal Robust Decision Trees Against Adversarial Examples Using Mixed-Integer Linear Programming

Daniël Vos and Sicco Verwer

11:10-11:30 MurTree: Optimal Classification Trees via Dynamic Programming and Search -- PDF

Emir Demirović, James Bailey, Rao Kotagiri, Christopher Leckie, Peter J. Stuckey, Anna Lukina, Emmanuel Hebrard and Jeffrey Chan

Session 5: Interplay between ML and optimization methods

11:30-11:50 Towards Unsatisfiable Core Learning for Chuffed -- PDF

Ronald van Driel and Neil Yorke-Smith

11:50-12:10 Divide and Learn: A Divide and Conquer Approach for Predict+Optimize -- PDF

Ali Guler, Emir Demirović, James Bailey, Jeffrey Chan, Christopher Leckie and Peter J. Stuckey

12:10-12:30 Interior Point Solving for LP-based prediction+optimisation -- PDF

Jayanta Mandi and Tias Guns

after-party in Gather.Town