Location visible to members
Come and learn about the machine learning challenges at Adform, and how to model football - and meet other people working with machine learning.
The purpose of these events is to inspire and broaden our knowledge within machine learning by sharing our expertise with each other - and at the same time getting to know each other a little bit better.
Maybe you have something to share next time?
• ~45 MIN. TALK
Machine learning at Adform
(by Vlad Sandulescu and Anders Nickelsen)
Vlad Sandulescu will present Adforms research group, the projects they are working on, the infrastructure, the toolbox and how they usually go from prototype to production. He will also talk about Adforms supply forecasting product which is powered by a time series forecasting model to predict future visits from users who match certain demographics.
Anders Nickelsen will talk about Adforms audience extension product that finds people with similar behavior and attributes among thousands of features. The product is used by advertisers to target e.g. people that are similar to visitors on their website. This product uses a combination of clustering and random forests to find similar people for thousands of groups of users through pipeline built in Spark and runs on Amazon EMR.
• INTRODUCING OURSELVES
Everybody gives an ultra short introduction to who they are or what they are working on - so that we have something to talk about in the break.
There will be beer and pizza ;)
• ~20 MIN. TALK
A review of modelling football
(by Lasse Bøhling)
This talk will take us through the history of modelling football, from the pioneers with pen and paper, to present day. We will see how to build a simple model using python and applying that information to generate probability distributions to beat the betting market.