Quantum Machine Learning with Application to Progressive Supranuclear Palsy Network Classification

Aug 3, 2026

Upload by Liam

Machine learning and quantum computing are combined to address computationally challenging classification tasks where large feature spaces slow classical methods and make kernel evaluations expensive. The approach reduces dimensionality with principal component analysis (PCA), then explores quantum learning on the PCA-reduced features. The framework is evaluated on a real clinical task: classifying Progressive Supranuclear Palsy (PSP) using functional connectivity network data. Results indicate improved performance with a variational quantum classifier achieving 86% accuracy and a quantum kernel estimator supporting a classical SVM, validated on IBM quantum simulators and hardware.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages41

Reading Time1 h 43 min

AccessFree to download

Introduction

Resting-state fMRI and brain network classification

Machine learning feature representation

Classical baselines: PCA and support vector machines

Kernel methods and computational bottlenecks

Quantum machine learning and variational quantum classification

Title: Quantum Machine Learning with Application to Progressive Supranuclear Palsy Network Classification Papri Saha* Department of Computer Science, Derozio Memorial College, Kolkata-700136, India ORCID: 0000-0001-9022-1947 Abstract: Machine learning and quantum computing are being progressively explored to shed light on possible computational approaches to deal with hitherto unsolvable problems. Classical methods for machine learning are ubiquitous in pattern recognition, with support vector machines (SVMs) being a prominent technique for network classification. However, there are limitations to the successful resolution of such classification instances when the input feature space becomes large, and the successive evaluation of so-called kernel functions becomes computationally exorbitant. The use of principal component analysis (PCA) substantially minimizes the dimensionality of feature space thereby enabling computational speed-ups of supervised learning: the creation of a classifier. Further, the application of quantum-based learning to the PCA reduced input feature space might offer an exponential speedup with fewer parameters. The present learning model is evaluated on a real clinical application: the diagnosis of Progressive Supranuclear Palsy (PSP) disorder. The results suggest that quantum machine learning has led to noticeable advancement and outperforms classical frameworks. The optimized variational quantum classifier classifies the PSP dataset with 86% accuracy as compared to conventional SVM. The other technique, a quantum kernel estimator, approximates the kernel function on the quantum machine and optimizes a classical SVM. In particular, we have demonstrated the successful application of the present model on both a quantum simulator and real chips of the IBM quantum platform. Keywords: quantum computing; support vector machine; principal component analysis; kernel estimation; Progressive Supranuclear Palsy. *Corresponding author information: Papri Saha, Rajarhat Road, P.O. - R-Gopalpur, Kolkata- 700136, India, (e-mail: [saha.papri@gmail.com](saha.papri@gmail.com); [pscompsc@dmc.ac.in](pscompsc@dmc.ac.in)). 1. Introduction The advent of “resting-state” functional magnetic resonance imaging (rs-fMRI) has allowed researchers to visualize large-scale cortical networks in the human brain by mapping the regions-of-interest (ROI) in terms of temporarily correlated low-frequency fluctuations offMRI signal that depends on the blood oxygen level [1, 2]. Many current studies have revealed that rs-fMRI-based network investigation is quite effective in identifying the unique signature of different neurological disorders [3-6]. Precisely, the functional brain network is a graph data structure with nodes representing the brain ROI and edges denoting the strength of the connection between those ROIs [7] . Therefore, the state-of-the-art diagnosis of any neurological disorder could emerge as a network classification problem. Imperatively, the primary task of such classification problems is the representation learning of graph-structured data. In particular, machine learning relates the feature space of factual data to discover generalized forms and intuitions with the aid of algorithms and statistical models without absolute instructions [8] . The intersection between machine learning (ML) and quantum computing has attracted significant interest in recent years [9-11] and several newly proposed quantum machine learning (QML) methods have been introduced to solve several real-life problems [12, 13] . The general approach assumes the initial problem of supervised learning: the creation of a classifier, where the network is presented with a labelled dataset 􀟯 = 􀜶 ∪ 􀜵 =((􀝔1, 􀝕1),(􀝔2, 􀝕2),…,(􀝔􀯠 , 􀝕􀯠), ) ⊂ 􀜴 􀯗 × {0, 1,…, 􀜿} . The training algorithm only considers the labels of the training data 􀜶 . The objective is to formulate a map on the test set 􀜵 → 􀜥 where 􀜥 = {0, 1,…, 􀜿}, such that it settles with high pr

FAQs

Why does the method use PCA before applying quantum machine learning?

PCA reduces the dimensionality of the feature space, which speeds up supervised learning and limits the computational cost associated with kernel evaluations.

What classical approach is used as a baseline for PSP network classification?

Support vector machines (SVMs) are used to construct a binary labeling function for distinguishing PSP patients from controls.

How does the quantum approach improve classification compared with classical SVM?

A variational quantum classifier and a quantum kernel estimator are applied to the PCA-reduced features and kernel estimation process, producing better results; the variational classifier reaches 86% accuracy on the PSP dataset.

Aug 3, 2026

Upload by Liam