QEM-Bench: Benchmarking Learning-based Quantum Error Mitigation and QEMFormer as a Multi-ranged Context Learning Baseline

Tianyi Bao, Ruizhe Zhong, Xinyu Ye, Yehui Tang, Junchi Yan

Abstract: Quantum Error Mitigation (QEM) has emerged as a pivotal technique for enhancing the reliability of noisy quantum devices in the *Noisy Intermediate-Scale Quantum* (NISQ) era. Recently, machine learning (ML)-based QEM approaches have demonstrated strong generalization capabilities without sampling overheads compared to conventional methods. However, evaluating these techniques is often hindered by a lack of standardized datasets and inconsistent experimental settings across different studies. In this work, we present **QEM-Bench**, a comprehensive benchmark suite of *twenty-two* datasets covering diverse circuit types and noise profiles, which provides a unified platform for comparing and advancing ML-based QEM methods. We further propose a refined ML-based QEM pipeline **QEMFormer**, which leverages a feature encoder that preserves local, global, and topological information, along with a two-branch model that captures short-range and long-range dependencies within the circuit. Empirical evaluations on QEM-Bench illustrate the superior performance of QEMFormer over existing baselines, underscoring the potential of integrated ML-QEM strategies.

Lay Summary: Quantum error mitigation (QEM) is crucial for enhancing the reliability of noisy NISQ devices; however, current ML-based QEM methods lack consistent benchmarks and employ heterogeneous evaluation protocols. To address this gap, we propose QEM-Bench, a comprehensive suite of twenty-two datasets that capture diverse circuit structures and noise regimes. We also introduce QEMFormer, an ML-driven framework that employs a multi-scale feature encoder to integrate local, global, and topological information, alongside a dual-branch architecture for modeling both short- and long-range circuit dependencies. Our empirical studies on QEM-Bench demonstrate that QEMFormer significantly outperforms existing baselines, establishing a unified standard for ML-based QEM research and paving the way for more accurate and scalable quantum computations.

Primary Area: Applications->Chemistry, Physics, and Earth Sciences

Keywords: Quantum Error Mitigation, Dataset Benchmark, Graph Learning

Submission Number: 9097