The development of solutions for the prediction of financial time series is considered a rather difficult problem, due to the many complex features frequently present in these series (irregularities, volatility, trends and noise). Many different paradigms have been studied for the development of prediction models able to determine the future behavior of a given phenomenon, or a time series, based on its past and present data. However, when the time series are from financial markets, a dilemma arises from some models for financial time series, known as random walk dilemma, where the predictions generated by such models show a characteristic one step delay with respect to the original time series data.
In this competition, the contestants are provided with ten (10) financial time series, each consisting of two hundred (200) points, and their objective is to obtain the next ten points for each series.
More details at http://www.gm2.com.br/ftspc
The objective is to generate a robust symbolic regression model that relates an expensive but noisy lab data of a chemical composition (output) to 57 cheap process measurements, such as temperatures, pressures, and flows (inputs). The selected equation has to include the most sensitive inputs relative to the output, i.e. some form of variables selection is recommended. If accepted by process engineers, the proposed symbolic regression solution could be implemented in a chemical process monitoring system.
The case is based on data from a real industrial application at Dow Chemical.
This competition links two interesting domains closely related to EvoStar: computational intelligence in games and optimisation under uncertainty. It is also very near to a situation often described in TV and sports magazines: the tuning of a car during the training days before the race when mechanics and pilots work on the car setup to find the one which will result in the best performance. The goal is to build an evolutionary (or related optimisation) algorithm that can replace the team of mechanics and pilots and can find the best car setup (e.g., gear ratio, wing area and angle, spring setup) on a given track.
A first version of this competition was held at GECCO 2009. However, we implement some changes here to bring it nearer to the real-world car setup optimisation process. Nevertheless, the main challenge remains that short evaluations might be noisy (the car drives only few meters before being evaluated) and that the optimisation algorithm itself has to determine how to use the available fixed time resources (1 million game ticks and no more than 2 hours of CPU time).
Deadline for submissions: April 1st.
More details at http://cig.dei.polimi.it/?page_id=103
Markus Kemmerling, Mike Preuss - TU Dortmund
The Mario AI Championship is the continuation of last year's very successful Mario AI Competition. The competition consists of three tracks. In the Gameplay track you submit an agent whose goal it is to survive as many levels of a modified version of Super Mario Bros as possible. In the Learning track, competitors submit agents that learn online to score as highly as possible on a specific level. The Level Generation sees competitors submit algorithms that generate as fun levels as possible for individual players. In all three tracks, we expect many of the competitors to submit entries based on evolutionary computation and similar bio-inspired algorithms, but other approaches are indeed possible and permitted.
Deadline for submissions: April 1st.
More details at http://www.marioai.org
Level Generation track: Noor Shaker, Julian Togelius and Georgios Yannakakis
This competition is open to everyone who enjoys using Evolutionary Computation to create art.
Deadline for submissions: April 7th.
More details at http://talpanet.org.uk/evocompetitions