Author ORCID Identifier

https://orcid.org/0009-0007-6185-5781

Semester

Summer

Date of Graduation

2026

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Lane Department of Computer Science and Electrical Engineering

Committee Chair

Piotr Wojciechowski

Committee Member

K. Subramani

Committee Member

Thomas Devine

Abstract

Combinatorial optimization problems (COPs) require searching over a finite solution space subject to constraints, with the goal of satisfying an objective function. They arise in operations research, scheduling, resource allocation, circuit design, and many other fields. Many problems in combinatorial optimization (including satisfiability and network design) are NP-hard. Traditionally, researchers have built approximate solvers that return near- optimal solutions efficiently by developing increasingly sophisticated heuristics and meta- heuristics. Deep learning has provided new opportunities for improving combinatorial solvers by leveraging neural guidance to prune the search space. Traditional neural networks have distinct drawbacks in this context: separate training and inference phases require significant computational resources and limit generalizability to unseen problem instances.

This thesis addresses the challenge by leveraging large-scale parallelization and data- less neural networks (dNNs), an emerging player in the field. The dataless approach has made real-time optimization possible without prior training. This advancement has substantially changed the memory needed to use a neural combinatorial solver. In this work, we have designed, implemented, and evaluated four dNNs for separate problems in satisfiability and network optimization. Experimental results show competitive or superior performance against commercial solvers, with improved scalability. We present empirical results for our implementations of the dataless approach.

Download

Share

COinS