Talk by Matthew T. Pratola (DSAS colloquium)
Room: 248
Title: Optimal Design Emulators and Near-Optimal Designs
Abstract:
Statistical design of experiments is a fundamental topic in applied statistics with a long history. Yet its application is often limited by the complexity and costliness of constructing experimental designs in the first place. For example, in optimal design, constructing a designed experiment involves searching the high-dimensional input space - a computationally expensive procedure that only guarantees a locally optimal solution. This is a hard problem that, typically, can only be “simplified” by changing the optimality criterion to be based on a simpler model. In this work, we introduce two novel approaches to the challenging design problem by taking a probabilistic perspective. In the first approach, a generative process is specified from which stochastic design realizations can be drawn. In the second approach, we introduce the concept of near-optimality, and describe an importance sampling algorithm to sample near-optimal designs.