Industrial Track Invited Speakers

  • Georges Hébrail
  • Dr. Georges Hébrail

  • Senior Researcher
  • EDF Lab, France
  • Title : Making smart metering smarter by applying data analytics

    Abstract : New data is being collected from electric smart meters which are deployed in many countries. Electric power meters measure and transmit to a central information system electric power consumption from every individual household or enterprise. The sampling rate may vary from 10 minutes to 24 hours and the latency to reach the central information system may vary from a few minutes to 24h. This generates a large amount of – possibly streaming – data if we consider customers from an entire country (ex. 35 millions in France). This data is collected firstly for billing purposes but can be processed with data analytics tools with several other goals. The first part of the talk will recall the structure of electric power smart metering data and review the different applications which are considered today for applying data analytics to such data. In a second part of the talk, we will focus on a specific problem: spatio-temporal estimation of aggregated electric power consumption from incomplete metering data.

    PDF Presentation: Making smart metering smarter by applying data analytics

    Bio : Georges Hébrail is a senior researcher at EDF Lab, the research centre of Electricité de France, one of the world’s leading electric utility. His background is in Business Intelligence covering many aspects from data storage and querying to data analytics. From 2002 to 2010, he was a professor of computer science at Telecom ParisTech, teaching and doing research in the field of information systems and business intelligence, with a focus on time series management, stream processing and mining. His current research interest is on distributed and privacy-preserving data mining on electric power related data.

  • Website
  • Alexandre Cotarmanac'h
  • Alexandre Cotarmanac’h

  • VP Platform & Distribution
  • Twenga, France
  • Title : Ads that matter

    Abstract : The advent of realtime bidding and online ad-exchanges has created a new and fast-growing competitive marketplace. In this new setting, media-buyers can make fine-grained decisions for each of the impressions being auctioned taking into account information from the context, the user and his/her past behavior. This new landscape is particularly interesting for online e-commerce players where user actions can also be measured online and thus allow for a complete measure of return on ad-spend.
    Despite those benefits, new challenges need to be addressed such as :
    – the design of a realtime bidding architecture handling high volumes of queries at low latencies,
    – the exploration of a sparse and volatile high-dimensional space
    – as well as several statistical modeling problems (e.g. pricing, offer and creative selection).
    In this talk, I will present an approach to realtime media buying for online e-commerce from our experience working in the field. I will review the aforementioned challenges and discuss open problems for serving ads that matter.

    Bio : Alexandre Cotarmanac’h is Vice-President Distribution & Platform for Twenga.
    Twenga is a services and solutions provider generating high value-added leads to online merchants that was founded in 2006.
    Originally hired to help launch Twenga’s second generation search engine and to manage the optimization of revenue, he launched in 2011 the affinitAD line of business and Twenga’s publisher network. Thanks to the advanced contextual analysis which allows for targeting the right audience according to their desire to buy e-commerce goods whilst keeping in line with the content offered, affinitAD brings Twenga’s e-commerce expertise to web publishers. Alexandre also oversees Twenga’s merchant programme and strives to offer Twenga’s merchants new services and solutions to improve their acquisition of customers.
    With over 14 years of experience, Alexandre has held a succession of increasingly responsible positions focusing on advertising and web development. Prior to joining Twenga, he was responsible for the development of Search and Advertising at Orange.
    Alexandre graduated from Ecole polytechnique.

  • Arthur von Eschen
  • Arthur von Eschen

  • Senior Director of Game Analytics
  • Activision Publishing Inc, USA
  • Title : Machine Learning and Data Mining in Call of Duty

    Abstract : Data science is relatively new to the video game industry, but it has quickly emerged as one of the main resources for ensuring game quality. At Activision, we leverage data science to analyze the behavior of our games and our players to improve in-game algorithms and the player experience. We use machine learning and data mining techniques to influence creative decisions and help inform the game design process. We also build analytic services that support the game in real-time; one example is a cheating detection system which is very similar to fraud detection systems used for credit cards and insurance. This talk will focus on our data science work for Call of Duty, one of the bestselling video games in the world.

    Bio : Arthur Von Eschen is Senior Director of Game Analytics at Activision. He and his team are responsible for analytics work that supports video game design on franchises such as Call of Duty and Skylanders. In addition to holding a PhD in Operations Research, Arthur has over 15 years of experience in analytics consulting and R&D with the U.S. Fortune 500. His work has spanned across industries such as banking, financial services, insurance, retail, CPG and now interactive entertainment (video games). Prior to Activision he worked at Fair Isaac Corporation (FICO). Before FICO he ran his own analytics consulting firm for six years.

  • Mike Bodkin
  • Mike Bodkin

  • VP Computational Chemistry and Cheminformatics
  • Evotec Ltd., UK
  • Title : Algorithms, Evolution and Network-Based Approaches in Molecular Discovery

    Abstract : Drug research generates huge quantities of data around targets, compounds and their effects. Network modelling can be used to describe such relationships with the aim to couple our understanding of disease networks with the changes in small molecule properties. This talk will build off of the data that is routinely captured in drug discovery and describe the methods and tools that we have developed for compound design using predictive modelling, evolutionary algorithms and network-based mining.

    PDF Presentation: Algorithms, Evolution and Network-Based Approaches in Molecular Discovery

    Bio : Mike did his PhD in protein de-novo design for Nobel laureate sir James Black before taking up a fellowship in computational drug design at Cambridge University. He moved to AstraZeneca as a computational chemist before joining Eli Lilly in 2000. Since 2003 he was head of the computational drug discovery group at Lilly but recently jumped ship to Evotec to work as the VP for computational chemistry and cheminformatics. His research aims are to continue to develop new algorithms and software in the fields of drug discovery and systems informatics and to deliver and apply current and novel methods as tools for use in drug research.