Amir Sani
I am a Machine Learning (Computer Science) PhD student at the University of Lille 1, supervised by Rémi Munos and Alessandro Lazaric. I am part of the SequeL Team at INRIA-Lille Nord-Europe.

My research interests are in
  • Risk averse stochastic and adversarial online learning (sequential decision making) algorithms and their application to asset (resource) management; with specific interest in Risk Sensitive Learning with Expert Advice and Online (Convex/Non-Convex) Optimization
    • With specific applications to
      • Online portfolio selection
      • Portfolio replication
      • Hedging with portfolios
      • Crypto-currency mining allocation (portfolio allocation with probabilistic rewards and no loss in resources due to portfolio losses)
  • Risk averse multi-arm bandit algorithms and their application to
    • Trade execution on the limit order book
    • Selecting policies/models/algorithms in partially observable sequential decision making environments
  • Model-free/Nonparametric (algorithms entirely dependent on the data with no model based assumptions) bootstrapping algorithms for dependent time series