Poster Title: Viability of Quantum Variational Classifiers for large and complex datasets

This study explores the potential of Quantum Variational Classifiers (QVCs) to process large datasets by comparing different ansatz configurations on benchmark datasets: Iris, Kinematics Motion, and Pulsar Classification. Utilising QVCs with RealAmplitudes, EfficientSU2, TwoLocal, and other ansatz types, I examined training times, accuracy, and scalability, comparing results against classical Support Vector Classifiers (SVCs). While SVCs consistently achieved high accuracy (e.g., 97% on Pulsar), QVCs demonstrated competitive performance on simpler datasets, with variations as dataset complexity increased. EfficientSU2, the most expressive ansatz, yielded improved flexibility and accuracy, though at the cost of increased training duration, while simpler ansatzes like TwoLocal balanced efficiency with moderate expressiveness, showing potential for feature-limited datasets. These preliminary findings highlight the computational trade-offs inherent in QVC architectures and suggest areas for optimisation as quantum machine learning progresses toward real-world applications, offering insights into overcoming current scalability challenges.