Research Papers

A Design Preference Elicitation Query as an Optimization Process

[+] Author and Article Information
Yi Ren1

Panos Y. Papalambros

Department of Mechanical Engineering,  University of Michigan, Ann Arbor, MI 48109pyp@umich.edu





Corresponding author.

J. Mech. Des 133(11), 111004 (Nov 11, 2011) (9 pages) doi:10.1115/1.4005104 History: Received March 22, 2011; Revised September 13, 2011; Published November 11, 2011; Online November 11, 2011

We seek to elicit individual design preferences through human-computer interaction. During an iteration of the interactive session, the computer queries the subject by presenting a set of designs from which the subject must make a choice. The computer uses this choice feedback and creates the next set of designs using knowledge accumulated from previous choices. Under the hypothesis that human responses are deterministic, we discuss how query schemes in the elicitation task can be viewed mathematically as learning or optimization algorithms. Two query schemes are defined. Query type 1 considers the subject’s binary choices as definite preferences, i.e., only preferred designs are chosen, while others are skipped; query type 2 treats choices as comparisons among a set, i.e., preferred designs are chosen relative to those in the current set but may be dropped in future iterations. We show that query type 1 can be considered as an active learning problem, while type 2 as a “black-box” optimization problem. This paper concentrates on query type 2. Two algorithms based on support vector machine and efficient global optimization search are presented and discussed. Early user tests for vehicle exterior styling preference elicitation are also presented.

Copyright © 2011 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

EGO operation concept

Grahic Jump Location
Figure 2

SVM search scattering and classification

Grahic Jump Location
Figure 3

SVM search design space reduction

Grahic Jump Location
Figure 4

Proposed SVM search algorithm

Grahic Jump Location
Figure 5

SVM search algorithm: scatter subroutine

Grahic Jump Location
Figure 6

Proposed EGO search algorithm

Grahic Jump Location
Figure 7

Comparison between EGO search and SVM search

Grahic Jump Location
Figure 8

Online human-computer interaction interface and data visualization. (a): The interactive environment3 allows the user to zoom, pan, rotate each design, and updates the guesses once the user hits the “Next” button; (b): The data visualization window4 has all user tests listed at the top, number of iterations in the middle, and the cumulated data at the bottom. The dotted curve represents the target design, the highlighted dark curve(s) represent the preferred design at this point, and the rest all not preferred.

Grahic Jump Location
Figure 9

Normalized Euclidean distance from each sampled design to the target in each test. The circled design is the one that submitted by the user, while the triangle design has the lowest Euclidean distance to the target.

Grahic Jump Location
Figure 10

Visual comparison between user test results and the targets from side and perspective views

Grahic Jump Location
Figure 11

Visual comparison between samples in the last iteration and the target. Data generated from Test0 on4 . Euclidean distances to the target are listed under the designs.

Grahic Jump Location
Figure 12

Designs labeled as preferred in the first four iterations, compared with the target and the one with the minimum Euclidean distance within all samples. Euclidean distances to the target are listed under the designs.

Grahic Jump Location
Figure 13

Features that capture the roof design

Grahic Jump Location
Figure 14

Samples from data Test0 in the feature and Euclidean space. Multidimensional scaling is applied to both measures to create 2D visualization of the data



Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In