Speaker
Dr Alan King,
Room “Sala Seminari” - Abacus Building (U14)
Doob Decomposition for AI-Driven Scenario Generation
Abstract
This talk discusses our experience in turning forecasts from Machine Learning or Time Series Foundation Model forecasts into a scenario generator. We start from the Doob decomposition: any adapted, integrable process (X_t) splits uniquely into a predictable forecast (A_t) and a martingale surprise (M_t). Any off-the-shelf forecaster can then serve as an estimator of (A_t), provided its output is genuinely predictable and nothing more. That “nothing more” is the hard part. We give a nonparametric three-step estimator for (A_t) and (M_t). The tested martingale increments then feed a scenario tree with an exact, node-by-node zero-sum constraint. We demonstrate the pipeline in a UK wind capacity-factor case study and close with what the theory does not yet cover.
Short bio
Alan King had a long and distinguished career at IBM Research in Yorktown Heights, New York.
At IBM, Alan participated in many research programs, including neural network proxy models, cryptocurrencies, and massively parallel computing, as manager, senior software engineer, lead consultant, university relations, and research staff advisor to senior leadership.
His research interest concerns the modeling and solution of stochastic programs, a branch of decision-making under uncertainty that applies to planning and operations. His contributions in this field include analysis of variance for solutions, software for solvers and tools, and modeling methodology.
Today, Alan is exploring how time series foundation models can be used to represent probability distributions in stochastic programs, with applications in energy and finance.
contact person for this Seminar: enza.messina@unimib.it