Distributed and Secure Kernel-Based Quantum Machine Learning
Authors: Arjhun Swaminathan, Mete Akgün
TMLR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The proposed architecture is validated using IBM s Qiskit Aer Simulator on various public datasets. Our methodology formalizes a robust framework that leverages quantum teleportation to enable secure and distributed kernel learning. The proposed architecture is validated using IBM s Qiskit Aer Simulator on various public datasets. |
| Researcher Affiliation | Academia | Arjhun Swaminathan1,2,* EMAIL Mete Akgün1,2 EMAIL 1Medical Data Privacy and Privacy-preserving Machine Learning (MDPPML), University of Tübingen 2Institute for Bioinformatics and Medical Informatics (IBMI), University of Tübingen *Corresponding Author |
| Pseudocode | No | The paper describes the protocol in Section 6.2 'Protocol Description' and provides a quantum circuit diagram in Figure 2, but it does not contain structured pseudocode or an algorithm block. |
| Open Source Code | Yes | Our code and results are available at the following URL: https://github.com/mdppml/ distributed-secure-kernel-based-QML. |
| Open Datasets | Yes | The datasets used for this analysis are widely used and publicly available. These include the Wine dataset (178 samples, 13 features) (Asuncion et al., 2007), the Parkinson s disease dataset (197 samples, 23 features) (Sakar et al., 2019), and the Framingham Heart Study dataset (4238 samples, 15 features) (Bhardwaj, 2022). ... We used a subset of the Digits dataset containing 100 samples (Pedregosa et al., 2011). |
| Dataset Splits | Yes | All SVM training and evaluation were performed using stratified 5-fold cross-validation to ensure unbiased accuracy metrics. |
| Hardware Specification | Yes | All the proof-of-concept experiments in our evaluation were carried out using classical computing resources in a High-Performance Computing (HPC) cluster. Each node within this HPC environment was equipped with an Intel XEON CPU E5-2650 v4, complemented by 256 GB of memory and 2 TB of SSD storage capacity. |
| Software Dependencies | No | The paper mentions "IBM s Qiskit Aer Simulator" but does not provide a specific version number for it or any other key software components. |
| Experiment Setup | Yes | Kernel-based training was performed using SVM for all datasets, and PCA was applied to the binary datasets (Parkinson s and Framingham Heart Study) to reduce dimensionality. All experiments were conducted with 1024 shots of the quantum circuit to ensure reliable accuracy. Here, shots refers to the number of times, p, a circuit is repeated. ... We employed three noise models: No Noise... Noise Level 1... Noise Level 2... We varied the number of shots, specifically using 128, 256, 512, and 1024 shots... |