Files in this item

FilesDescriptionFormat

application/pdf

application/pdfGE-THESIS-2018.pdf (4MB)
(no description provided)PDF

Description

Title:Computational creativity applications in engineering
Author(s):Ge, Xiou
Advisor(s):Varshney, Lav R.
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:M.S.
Genre:Thesis
Subject(s):Computational Creativity, Engineering
Abstract:We investigated and implemented two computational creativity applications in generating engineering processes and building material designs respectively. To synthesize engineering processes, we developed a full system that generates engaging Rube Goldberg Machine designs. We first describe the use of case-based reasoning (CBR) and an existing knowledge base to yield a combinatorial design space for experiments. We then apply automated planning techniques to generate experiment procedures. We further use functional modeling to represent the experiment devices and demonstrate how that representation enables the planner to generate a valid Rube Goldberg Machine. Finally, a semantic similarity metric is proposed to evaluate the quality of a generated chain of experiments. To discover concrete formulas as building materials with desired properties, we use a conditional variational autoencoder (CVAE), a type of semisupervised generative model. Our model is trained using open data from the UCI Machine Learning Repository joined with environmental impact data computed using a web-based tool. We demonstrate that the CVAE can design concrete formulas with lower emissions and natural resource usage while meeting design requirements. To ensure fair comparison between extant and generated formulas, we also train regression models to predict the environmental impacts and strength of discovered formulas. With these results, a construction engineer may create a formula that meets structural needs and best addresses local environmental concerns.
Issue Date:2018-12-10
Type:Thesis
URI:http://hdl.handle.net/2142/102507
Rights Information:Copyright 2018 Xiou Ge
Date Available in IDEALS:2019-02-06
Date Deposited:2018-12


This item appears in the following Collection(s)

Item Statistics