Optimal Learning for Efficient Experimentation in Nanotechnology and Biochemistry

Abstract

We have extended our work in optimal learning for continuous parameters, which previously was limited to searching among a small set of possible sample values, to amore general resampling approach that allows us to quickly find the best parameter values globally. We also developed a version of Peptide Optimization with Optimal Learning (POOL) that can be used with quantitative responses, rather than the binary responses assumed by the first version of the method. This greatly expands the set of applications to which POOL can be applied. We also developed a statistical method for characterizing the bias in phage display, and making predictions of activity based on phage display data. This method can be used on its own, and can also provide input to POOL. Finally, we have developed a series of tutorial materials, first in the form of a series of powerpoint presentations, and second as a book chapter, both geared toward materials scientists.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 22, 2015
Accession Number
AD1001807

Entities

People

  • Peter Frazier
  • Warren B. Powell

Organizations

  • Trustees of Princeton University

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Air Force Research Laboratories
  • Algorithms
  • Chemistry
  • Computational Science
  • Machine Learning
  • Materials
  • Materials Science
  • Military Research
  • Nanotechnology
  • Nonlinear Dynamics
  • Operations Research
  • Probability
  • Probability Distributions
  • Random Variables
  • Research Facilities
  • Statistical Analysis
  • Training

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Computer Science.
  • Systems Analysis and Design

Technology Areas

  • Biotechnology