PyData Amsterdam: Learning to Rank & Grouping Behaviour related to Music

PyData Amsterdam
PyData Amsterdam
Openbare groep
Locatieafbeelding van evenementslocatie

Wat we doen

Hi PyData folks,

We're back after the summer and it's time for a new meetup! Kindly hosted by GoDataDriven, we have two amazing talks lined up for you which are both part of (ongoing) research projects at the University of Amsterdam.

Program
18:15 Walk-in, pizzas and drinks!
19:00 Unbiased Learning to Rank from User Interactions by Harrie Oosterhuis
19:30 Small break
19:45 Using Overhead Video Capture to Analyse Grouping Behaviour of Dancers in a Silent Disco by Nelson Mooren
20:15 More time to mingle and drink

Around 21:30 the bar will close (and the building as well!)

Talks
### Unbiased Learning to Rank from User Interactions
Learning to rank provides methods for optimising ranking systems, enabling effective search and recommendation systems. Traditionally, these methods relied on annotated datasets i.e. relevance labels for query-document pairs provided by human judges. Over the years, the limitations of such datasets have become apparent, most importantly: they are expensive to create and do not necessarily reflect user preferences. Recently attention has mostly shifted to methods that learn from user interactions, as they more closely indicate user preferences. However, user interactions contain large amounts of noise and bias, learning while naively ignoring these biases can lead to detrimental results. Consequently, the current focus is on unbiased methods that can reliably learn from user interactions. In this talk I will describe the main approaches to unbiased learning to rank and discuss the most recent methods from the field.

About Harrie Oosterhuis
Harrie Oosterhuis (https://staff.fnwi.uva.nl/h.r.oosterhuis) is a 3rd year PhD student under supervision of Prof. dr. Maarten de Rijke at the University of Amsterdam. His main topic is learning to rank from user behaviour and he has publications at major IR conferences including CIKM, SIGIR, ECIR and WSDM. In addition he has completed multiple internships at Google Research & Brain in California, and worked as a visiting researcher at RMIT university in Melbourne during his PhD.

### Using Overhead Video Capture to Analyse Grouping Behaviour of Dancers in a Silent Disco
Silent-disco events present experimenters with an ecologically valid environment in which to study the interactions between music and social behaviour. This can be done in large groups of participants and at a relatively low cost compared to laboratory settings. I developed a toolbox to process overhead video data and to investigate grouping dynamics of dancers in a silent disco with multiple channels. Using these methods, I showed that differences in grouping behaviour are related to the different musical styles that people listened to.

About Nelson Mooren
Nelson is a neuroscientist turned software developer with an interest in information processing and computational modelling. The presented research was conducted at Music Cognition Group of the University of Amsterdam. Nelson currently works in the domain of fraud detection.