Program & How to Attend
How to join
The virtual venue (gather.town) has now officially opened and can be visited!
To attend the workshop via gather.town:
Click the ‘Enter here’ link on the following page https://ijcai-21.org/ijcai-virtual-space-access/
Log into gather.town using the same credentials you used to originally register for IJCAI.
On first login, this will open the gather.town service, where you can set up your profile and avatar.
Our workshop code is W09, held in room ‘Red 1’. During the day of the workshop, to more easily find the room, we recommend you use the following instructions:
In gather.town, click on the ‘Participants’ icon on the bottom left of the screen
Search for one of the organizers names (e.g., Karl Tuyls, Jesse Davis, Jan Van Haaren, Shayegan Omidshafiei), then click their name
Click ‘Locate on Map’
A black line indicating a path to their location will appear, which you can follow to the workshop room.
If you have any problems logging in to the gather.town service, finding the workshop room, technical difficulties, or any other questions, please contact one of the organizers (e.g., karltuyls@google.com, jan.vanhaaren@kuleuven.be, jesse.davis@kuleuven.be, somidshafiei@google.com).
For more detailed instructions (e.g., how to navigate, share screen, etc.), please check out the document below.
IJCAI Gathertown Guide.pdfProgram (August 17th, 2021)
The following details the workshop program.
The papers in each session will be presented in the order specified below. Please note that all extended abstracts are allotted 5 minutes (2 minutes for the talk, 3 minutes for Q&A), and full papers 10 minutes (5 minutes for the talk, 5 minutes for Q&A).
For your convenience, an optional Google Calendar has been set up here to help keep track of session times. Note that as the workshop itself is being held fully over gather.town, you should follow the instructions above to see how to attend/present.
Session 1 (Introductory Remarks) | 2:00 pm - 2:15 pm CEST / 8:00 am - 8:15 am EDT
Session 2 (Invited Talk) | 2:15 pm - 3:05 pm CEST / 8:15 am - 9:05 am EDT
Speaker: Stephanie Kovalchik (Zelus Analytics)
Title: Latent Style Allocation for Describing Serve Return Patterns in Professional Tennis
Abstract: Recently publicized tracking data for men's professional tennis has created the opportunity for the first detailed spatial analysis of the return impact characteristics of elite tennis players. Mixture models are an appealing model-based framework for spatial analysis in sport where latent variable discovery is often of primary interest. Although finite mixture models have the advantages of being highly interpretable and scalable, most implementations have the conditional spatial distribution within each latent subgroup as a standard parametric distribution. In this talk, I'll present a more flexible alternative that allows the conditional distribution in each latent category to be a mixed member of finite Gaussian mixtures. This model was motivated by efforts to describe common styles of return impact location of professional tennis players. In a fully Bayesian implementation in the Stan language, we apply the model to 142,803 return points played by 141 pro men's players between 2018 and 2020 and identify six unique impact styles on the first and second serve return.
30 Minute Break | 3:05 pm - 3:35 pm CEST / 9:05 am - 9:35 am EDT
Session 3 (Paper Talks) | 3:35 pm - 4:30 pm CEST / 9:35 am - 10:30 am EDT
[5min extended abstract] Tatsuya Yoshikawa, Kazushi Tsutsui, Kazuya Takeda and Keisuke Fujii. Extraction of swing motion contributing to prediction of shuttle drop position in badminton (video)
[10min full paper] Maaike Van Roy, Wen-Chi Yang, Luc De Raedt and Jesse Davis. Analyzing Learned Markov Decision Processes using Model Checking for Providing Tactical Advice in Professional Soccer (video)
[10min full paper] Lotte Bransen and Jesse Davis. Women's football analyzed: interpretable expected goals models for women
[10min full paper] Jan Van Haaren. "Why Would I Trust Your Numbers?" On the Explainability of Expected Values in Soccer (short video)
[10min full paper] Théophane Gregoir and Paul Garnier. Evaluating Football Player: from Live Camera to Deep Reinforcement Learning
[10min full paper] Adrià Arbués-Sangüesa, Adrián Martín, Paulino Granero, Coloma Ballester and Gloria Haro. Learning Football Body-Orientation as a Matter of Classification
Session 4 (Invited Talk) | 4:30 pm - 5:20 pm CEST / 10:30 am - 11:20 am EDT
Speaker: William Spearman (Liverpool Football Club)
Title: Pitch Control
30 Minute Break | 5:20 pm - 5:50 pm CEST / 11:20 am - 11:50 am EDT
Session 5 (Paper Talks) | 5:50 pm - 6:55 pm CEST / 11:50 am - 12:55 pm EDT
[10min full paper] Yaniv Oshrat, Matan Akrabi, Ariel Roth and Sarit Kraus. Finding the Right Partner for Doubles Matches: A Data-driven Study on Bughouse Chess
[10min full paper] Peter Xenopoulos, Bruno Coelho and Claudio Silva. Optimal Team Economic Decisions in Counter-Strike
[10min full paper] Guido Petri, Michael H. Stanley, Alec B. Hon, Alexander Dong, Peter Xenopoulos and Cláudio Silva. Bandit Modeling of Map Selection in Counter-Strike: Global Offensive
[10min full paper] David Radke, Daniel Radke, Tim Brecht and Alex Pawelczyk. Passing and Pressure Metrics in Ice Hockey (video)
[10min full paper] Konstantinos Pelechrinis and Kirk Goldsberry. The Anatomy of Corner 3s in the NBA: What makes them efficient, how are they generated and how can defenses respond? (video)
[10min full paper] Guande Wu, Shuya Zhao, Jianzhe Lin and Claudio Silva. Basketball GAN: Sportingly Acceptable Trajectory Prediction
[5min extended abstract] Preston Biro and Stephen Walker. A Reinforcement Learning Based Approach to Play Calling in Football
30 Minute Break | 6:55 pm - 7:25 pm CEST / 12:55 pm - 1:25 pm EDT
Session 6 (Panel Discussion) | 7:25 pm - 8:25 pm CEST / 1:25 pm - 2:25 pm EDT
Chris Brady (Sportsology)
Luke Bornn (Zelus Analytics)
Christina Chase (MIT Sports Lab)
Jesse Davis (KU Leuven)
Ian Graham (Liverpool Football Club)