Presented By O’Reilly and Cloudera
Make Data Work
September 11, 2018: Training & Tutorials
September 12–13, 2018: Keynotes & Sessions
New York, NY

Real-time machine intelligence in IndyCar and Tour de France

Yasuyuki Kataoka (NTT Innovation Institute, Inc.)
4:20pm–5:00pm Thursday, 09/13/2018
Data-driven business management, Strata Business Summit
Location: 1E 10/11 Level: Beginner
Secondary topics:  Transportation and Logistics
Average rating: ***..
(3.00, 4 ratings)

Who is this presentation for?

  • Data scientists

Prerequisite knowledge

  • A general understanding of machine learning

What you'll learn

  • Explore real examples of machine learning that delivers meaningful insights beyond a monitoring tool, including a noise reduction technique for wearable sensors, a spatiotemporal data prediction technique, and a deployment pattern for a real-time machine learning model

Description

With the invention of a vast variety of wearable sensors, the real-time sports analytics is being innovated more than ever by leveraging time-series and heterogenous data. The outcome of sports data analytics is utilized to provide real-time feedback to professional athletes and fans. While this feedback differs for each sport, one of the common challenges is noise reduction from wearable sensors in extreme conditions. But if you work on time series and heterogenous data captured in real time in sports or applications such as human monitoring, how do you effectively deploy machine learning models in real time? Further, how do you discover and deliver meaningful perspective beyond a simple real-time monitoring tool?

Yasuyuki Kataoka highlights various real-time machine learning models in both IndyCar and Tour de France, sharing real-time data processing architectures, machine learning models, and demonstrations that deliver meaningful insights for players and fans. Yasuyuki first details a real-time data quality assessment technique for IndyCar that enables 99.5% accuracy for noise reduction in the judgment of whether data is reliable or not. Yasuyuki then explores real-time machine learning models for both IndyCar and Tour de France that provide actionable insights for the players or entertaining aspects for the fans, including a cycling use case where machine learning uses only GPS trajectory data to predict muscle performance, which is the key performance factor in cycling sports, and offers an overview of a data analytics platform powered by IBM that successfully performs in the real competition. Along the way, Yasuyuki discusses the trade-off between latency and accuracy—a typical problem in real-time machine learning applications.

Photo of Yasuyuki Kataoka

Yasuyuki Kataoka

NTT Innovation Institute, Inc.

Yasuyuki Kataoka is a data scientist at NTT Innovation Institute, Inc. His primary interest is applied R&D in machine learning applications for time series and heterogeneous data such as vision, audio, text, and IoT sensor signals. This data science work spans various fields including automotive, sports, healthcare, and social media. Other areas of interest include robotics control such as self-driving car and drone systems. When not doing research activities, he likes to participate in hackathons, where he has won prizes in the automotive and healthcare industries. Yasuyuki is a PhD candidate in artificial intelligence at the University of Tokyo and holds an MS and BS in mechanical and system engineering from Tokyo Institute of Technology, where he graduated with valedictorian honors.