Cside 2018: Competitive statistical inference for differential equations 26th Nov 2018, University of Glasgow Mathematics and Statistics Building, Room 311B


Differential equations are a powerful way of providing an observed system a mathematical description. However, the act of inferring the parameters that govern the equations from noisy data, especially in a timely fashion, is challenging. The main objective of Cside is to assess in a competitive setting what inference methods in this area are currently the state of the art. Due to the multi-disciplinary nature of the research area, it should be of interest to biologists, computer scientists, mathematicians and statisticians.

The format for the event is as follows: Competitors can use whatever method they wish to infer the parameters of simulated data. Since the parameters that generated the data are known, we can quantify how well the method performed, subject to some assessment criteria. With many competitors working on the same problem, we can get an overview as to what methods perform best and in what fashion. This year we have three models to choose from; an Ordinary Differential Equation system, a Partial Differential Equation system and a Stochastic Differential Equation system.

Between now and the end of October, competitors should spend some time familiarising themselves with their chosen model. At the end of October, competitors will be given a dataset generated from their chosen model by the event organiser. You will have one week to infer the parameters of this dataset. At the end of November there will be a one day conference discussing the results of the challenge. The best 3 performing competitors for each model (9 total) will be invited to give an oral presentation on this day.

This is an excellent opportunity to test your methods, identify gaps in research and generate new ideas for developing future techniques. The invited speakers will have access to travel and accommodation support (capped) and there will be a prize for the best performing method for each model (3 total). Those interested in systems described by differential equations, who either do not work on parameter inference or are unable to compete, are more than welcome to attend the conference day as attendees-only. The conference day is a great platform for discussing ideas, and Cside aims to promote and facilitate multi-disciplinary collaboration. Conference attendees are encouraged to bring a poster showcasing their current work, and there will be a dedicated time-slot for poster presentation and networking.

Cside welcomes participants of all research backgrounds and levels of expertise. We would love to see a good amount of participation from early career researchers in order to support their personal development, so please don't be shy!

Full details of the event can be found at www.tinyurl.com/cside2018


First published: 3 September 2018