Research

I work on statistical learning theory with a particular focus on online learning: sequential prediction and aggregation of experts, bandit problems (stochastic, adversarial, sleeping, dueling), online convex optimisation, nonparametric regression, and applications to reinforcement learning and demand forecasting. My PhD thesis, supervised by Yannig Goude and Gilles Stoltz, focused on prediction of individual sequences.

Google Scholar  ·  ORCID

Publications

Bandits and online decision making

Online learning and aggregation

Nonparametric methods

Optimisation

Reinforcement learning

Counterfactual and off-policy learning

Forecasting and applications

PhD thesis

Software

  • Opera: Online Prediction by ExpeRt Aggregation. Pierre Gaillard, Yannig Goude. R package, 2016.
    Opera is an R package for prediction of time series based on online robust aggregation of a finite set of forecasts (machine learning method, statistical model, physical model, human expertise…). More formally, we consider a sequence of observation y(1),…,y(t) to be predicted element by element. At each time instance t, a finite set of experts provide prediction x(k,t) of the next observation y(t). Several methods are implemented to combine these expert forecasts according to their past performance (several loss functions are implemented to measure it). These combining methods satisfy robust finite time theoretical performance guarantees. We demonstrate on different examples from energy markets (electricity demand, electricity prices, solar and wind power time series) the interest of this approach both in terms of forecasting performance and time series analysis.

PhD Students and postdocs

PhD students currently supervised

Former PhD students and postdocs

School

Here, you can find some reports I wrote during my studies.