The NSF Workshop on Learning and Approximate Dynamic Programming is proposed to take place on April 8-10, 2002, in Playacar, Mexico. The technical focus of the workshop is to address one common cross-cutting problem: how do we do our best, using learning-based approaches or the equivalent, to develop general-purpose designs to try to maximize the sum of expected utility over future time, in a nonlinear stochastic environment, without just simulating the entire future?
An important goal of the workshop is to bring people in diverse areas to unify their efforts to tackle some important problems. The workshop is therefore highly interdisciplinary, and the invitees are world leaders in machine learning, control systems theory, and neural networks. The technical program is intended to create opportunities for participants to carry on discussions and to form teams for their future work.