Members of our SLiQ Labs research team will be making their way to New York City for AAAI-20 on February 7-12. The purpose of the AAAI conference is to promote research in artificial intelligence and scientific exchange among AI researchers, practitioners, scientists, and engineers in affiliated disciplines. AAAI-20 will have a diverse technical track, abstracts, poster sessions, invited speakers, tutorials, workshops, and exhibit and competition programs, all selected according to the highest reviewing standards.
Sportlogiq is thrilled to be a part of this important event, which provides us with a platform to showcase the groundbreaking technology we are working on. Mehrsan Javan, Sportlogiq CTO & co-founder, will be delivering the opening keynote address at the AI in Team Sports workshop. He will be speaking alongside other industry leaders, such as Yann LeCun (New York University and Facebook) and Susan Athey (Stanford University). Our SLiQ Labs Director of Predictive Analytics Research, Oliver Schulte, will also have a significant presence at AAAI-20, and will be presenting 2 papers at the conference. Below are some details on what you can expect to see from the SLiQ Labs team.
Title: Real-time Sport Analytics
Abstract: Sports analytics is about observing, understanding, describing, and predicting the game in an intelligent manner. In practice, this means designing a fully-automated, robust, end-to-end pipeline; from visual input, to player and group activities, to player and team evaluation, to planning. This talk will cover the components of such a system and provides examples of how Sportlogiq captures the data from videos and generates relevant insights in real-time for different stakeholders in the sports industry.
Read more about the AI in Team Sports workshop here.
Title: Valuing Sports Actions and Players with Inverse Reinforcement Learning
Abstract: We all know that teams want to score, but how do they try to get there? We use machine learning to get inside a team’s head to model what objectives they pursue to get to a goal. For example, do they care more about power plays or about breakaways? These are the questions that we try to answer with inverse reinforcement learning in our paper: Valuing Sports Actions and Players with Inverse Reinforcement Learning.
Title: Learning Contextualized Player Representations with a Variational Hierarchical Encoder
Abstract: One of the key questions in modelling sports is how players are alike and how they are different. We use deep learning to describe players by learned features so that players with similar features can be expected to act in similar ways. Building on the state-of-the-art, we introduce a powerful new deep model called the hierarchical variational auto-encoder to learn player features. Combining this way of describing players with data about who has the puck when, improves performance on a number of tasks, including player identification, predicting successful shots, and the final match outcome. To learn more, take a look at the paper: Learning Contextualized Player Representations with a Variational Hierarchical Encoder.
Send us an email at email@example.com to connect with us at AAAI-20. We hope to see you there!