Keyword Suggestions

WW1.05 - Optimal Learning with an Application to Characterizing Nano-Emulsion Stability 
April 22, 2014   9:45am - 10:00am

Scientists are often posed with the problem of finding a good choice of experimental parameters that optimize a particular quantity in the presence of ambiguity about the underlying system. This ambiguity arises from unknown physical constants, competing (and often poorly understood) mechanisms, and measurement or calibration error. Instead of an exhaustive and potentially expensive search over all such parameters, the method of Optimal Learning uses Bayesian statistics and heuristics to guide an experimenter through parameter space in an efficient manner, with the goal of learning about the underlying physical system through each experiment. We give an overview of the Optimal Learning method and the heuristics involved. We present how a meaningful baseline prior belief is elicited, using the domain knowledge of the experimenter. We then present a real-world application of the Optimal Learning technique to characterizing the stability of nano-emulsions stabilized by gold nanorods.

Average Rating: (1 Rating)
  Was great, surpassed expectations, and I would recommend this
  Was good, met expectations, and I would recommend this
  Was okay, met most expectations
  Was okay but did not meet expectations
  Was bad and I would not recommend this

Keynote Address
Panel Discussion - Different Approaches to Commercializing Materials Research
Business Challenges to Starting a Materials-Based Company
Fred Kavli Distinguished Lectureship in Nanoscience
Application of In-situ X-ray Absorption, Emission and Powder Diffraction Studies in Nanomaterials Research - From the Design of an In-situ Experiment to Data Analysis