Abstract

Quantum error mitigation (QEM) is essential for reliable near-term quantum computing, yet practical deployment requires balancing mitigation strength against runtime overhead under nonstationary noise. This work introduces GSC-QEMit, a telemetry-driven context–forecast–bandit framework that adap- tively selects mitigation intensity using unsupervised context discovery, uncertainty- aware Gaussian process forecasting, and a cost-aware contextual bandit. Eval- uated on benchmark quantum circuits under drifting noise in simulation, the approach improves logical fidelity while reducing unnecessary interven- tion cost. Complementing this systems-level contribution, we apply unsu- pervised machine learning to broadband cryogenic transient dielectric spec- troscopy data to analyze two-level system (TLS) noise mechanisms. Through clustering, dimensionality reduction, and spectral analysis, we recover invari- ant frequency structures and interference patterns directly from raw signals without explicit physical models. Together, these results demonstrate a uni- fied role for machine learning in enabling practical quantum computing, both through adaptive control of system behavior and data-driven understanding of underlying physical noise processes.

Publication Date

7-10-2026

Document Type

Thesis

Student Type

Graduate

Degree Name

Data Science (MS)

Department, Program, or Center

Software Engineering, Department of

College

Golisano College of Computing and Information Sciences

Advisor

Daniel Krutz

Advisor/Committee Member

Travis Desell

Campus

RIT – Main Campus

Download

Share

COinS