Challenges 2019
This year's challenges were organized and supervised at ENS by Tomas Anglès, Louis Thiry, Roberto Leonarduzzi and John Zarka. The organization of these data challenges is supported by the CFM Chair at the Ecole Normale Supérieure, and by the Fondation des Sciences Mathématiques de Paris.
In this first session, the following 7 challenges are presented:
- " Predicting daily movements in US equities " presented by Éric Lebigot of Capital Fund Management. The aim of the challenge is to predict the sign of US stock returns over a 30-minute period at the end of the day, based on their historical returns over 5-minute windows at the beginning of the day.
- " Detecting breast cancer metastases " presented by Charlie Saillard from Owkin. The aim of the challenge is to determine the presence of lymph node metastases in histological section images of breast cancer patients. This challenge is " weakly supervised ", the presence or absence of metastases being indicated locally for some images, but only globally at the scale of the whole image for most sections.
- " Prediction of dynamic electricity profiles presented by Pierre Cauchois from Enedis. The aim of the challenge is to estimate 7 series of dynamic electricity profiles representing the consumption patterns of different mass market customer categories, based on energy and weather data.
- " Sharpe ratio prediction of mixtures of quantitative strategies " presented by Stefan Duprey of Napoléon Crypto. The aim of the challenge is to predict the Sharpe ratio, an indicator of risk-adjusted returns, of specified combinations of 7 quantitative strategies over a period of 5 business days, based on the 21-business-day history of returns of these 7 strategies as well as 3 relevant financial indicators.
- " Prediction of slow brain wave activity during deep sleep " presented by Valentin Thorey from Dreem. The aim of the challenge is to predict, from 10 seconds of EEG signals of a slow brain wave during deep sleep, as well as various indicators of sleep quality up to that point, whether another slow wave of low or high amplitude will follow.
- " Predicting expected response to pharmaceutical questions " presented by Emmanuel Bilbault from Posos. The aim of the challenge is to categorize pharmaceutical questions according to the type of answer expected.
- " Solving the Rubik's cube 2x2x2 presented by Julien Peyras from LumenAI. The aim of the challenge is to predict, from a certain initial configuration of a 2x2x2 Rubik's cube, the minimum number of moves required to reach the solution.
