Showing 1–2 of 2 results for author: Anguera, X

The Zero Resource Speech Challenge 2017

Authors: Ewan Dunbar, Xuan Nga Cao, Juan Benjumea, Julien Karadayi, Mathieu Bernard, Laurent Besacier, Xavier Anguera, Emmanuel Dupoux

Abstract: We describe a new challenge aimed at discovering subword and word units from raw speech. This challenge is the followup to the Zero Resource Speech Challenge 2015. It aims at constructing systems that generalize across languages and adapt to new speakers. The design features and evaluation metrics of the challenge are presented and the results of seventeen models are discussed. We describe a new challenge aimed at discovering subword and word units from raw speech. This challenge is the followup to the Zero Resource Speech Challenge 2015. It aims at constructing systems that generalize across languages and adapt to new speakers. The design features and evaluation metrics of the challenge are presented and the results of seventeen models are discussed. △ Less

Submitted 12 December, 2017; originally announced December 2017.

Comments: IEEE ASRU (Automatic Speech Recognition and Understanding) 2017. Okinawa, Japan

Automatic Extraction of the Passing Strategies of Soccer Teams

Authors: Laszlo Gyarmati, Xavier Anguera

Abstract: Technology offers new ways to measure the locations of the players and of the ball in sports. This translates to the trajectories the ball takes on the field as a result of the tactics the team applies. The challenge professionals in soccer are facing is to take the reverse path: given the trajectories of the ball is it possible to infer the underlying strategy/tactic of a team? We propose a metho… ▽ More Technology offers new ways to measure the locations of the players and of the ball in sports. This translates to the trajectories the ball takes on the field as a result of the tactics the team applies. The challenge professionals in soccer are facing is to take the reverse path: given the trajectories of the ball is it possible to infer the underlying strategy/tactic of a team? We propose a method based on Dynamic Time War** to reveal the tactics of a team through the analysis of repeating series of events. Based on the analysis of an entire season, we derive insights such as passing strategies for maintaining ball possession or counter attacks, and passing styles with a focus on the team or on the capabilities of the individual players. △ Less

Submitted 10 August, 2015; originally announced August 2015.

Comments: 2015 KDD Workshop on Large-Scale Sports Analytics