Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Practical Quantum-Classical Feature Fusion for complex data Classification

Azadeh Alavi, Fatemeh Kouchmeshki, Abdolrahman Alavi

TL;DR

This work addresses the challenge of integrating quantum feature maps with classical classifiers for complex data. It reframes hybrid quantum-classical learning as a multimodal fusion problem and introduces a cross-attention mid-fusion architecture that lets a classical pathway query quantum-derived tokens. Across five benchmark datasets, pure quantum and simple hybrid approaches underperform due to measurement bottlenecks, while cross-attention mid fusion offers consistent gains, especially on high-dimensional tasks. The results suggest that quantum information is most valuable when integrated through principled multimodal interaction, guiding practical design for NISQ-era quantum-classical systems.

Abstract

Hybrid quantum and classical learning aims to couple quantum feature maps with the robustness of classical neural networks, yet most architectures treat the quantum circuit as an isolated feature extractor and merge its measurements with classical representations by direct concatenation. This neglects that the quantum and classical branches constitute distinct computational modalities and limits reliable performance on complex, high dimensional tabular and semi structured data, including remote sensing, environmental monitoring, and medical diagnostics. We present a multimodal formulation of hybrid learning and propose a cross attention mid fusion architecture in which a classical representation queries quantum derived feature tokens through an attention block with residual connectivity. The quantum branch is kept within practical NISQ budgets and uses up to nine qubits. We evaluate on Wine, Breast Cancer, Forest CoverType, FashionMNIST, and SteelPlatesFaults, comparing a quantum only model, a classical baseline, residual hybrid models, and the proposed mid fusion model under a consistent protocol. Pure quantum and standard hybrid designs underperform due to measurement induced information loss, while cross attention mid fusion is consistently competitive and improves performance on the more complex datasets in most cases. These findings suggest that quantum derived information becomes most valuable when integrated through principled multimodal fusion rather than used in isolation or loosely appended to classical features.

Practical Quantum-Classical Feature Fusion for complex data Classification

TL;DR

This work addresses the challenge of integrating quantum feature maps with classical classifiers for complex data. It reframes hybrid quantum-classical learning as a multimodal fusion problem and introduces a cross-attention mid-fusion architecture that lets a classical pathway query quantum-derived tokens. Across five benchmark datasets, pure quantum and simple hybrid approaches underperform due to measurement bottlenecks, while cross-attention mid fusion offers consistent gains, especially on high-dimensional tasks. The results suggest that quantum information is most valuable when integrated through principled multimodal interaction, guiding practical design for NISQ-era quantum-classical systems.

Abstract

Hybrid quantum and classical learning aims to couple quantum feature maps with the robustness of classical neural networks, yet most architectures treat the quantum circuit as an isolated feature extractor and merge its measurements with classical representations by direct concatenation. This neglects that the quantum and classical branches constitute distinct computational modalities and limits reliable performance on complex, high dimensional tabular and semi structured data, including remote sensing, environmental monitoring, and medical diagnostics. We present a multimodal formulation of hybrid learning and propose a cross attention mid fusion architecture in which a classical representation queries quantum derived feature tokens through an attention block with residual connectivity. The quantum branch is kept within practical NISQ budgets and uses up to nine qubits. We evaluate on Wine, Breast Cancer, Forest CoverType, FashionMNIST, and SteelPlatesFaults, comparing a quantum only model, a classical baseline, residual hybrid models, and the proposed mid fusion model under a consistent protocol. Pure quantum and standard hybrid designs underperform due to measurement induced information loss, while cross attention mid fusion is consistently competitive and improves performance on the more complex datasets in most cases. These findings suggest that quantum derived information becomes most valuable when integrated through principled multimodal fusion rather than used in isolation or loosely appended to classical features.
Paper Structure (42 sections, 18 equations, 3 figures, 7 tables)

This paper contains 42 sections, 18 equations, 3 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Methodology
  4. Datasets and learning tasks
  5. Wine.
  6. Breast Cancer Wisconsin (Diagnostic).
  7. Covertype.
  8. Fashion-MNIST.
  9. Steel Plates Faults.
  10. Cross-validation, monitor split, and leakage control.
  11. Datasets and Experimental Setup
  12. Problem Setting and Notation
  13. Preprocessing and Dimensionality Control
  14. Standardization
  15. Principal Component Projections
  16. ...and 27 more sections

Figures (3)

  • Figure 1: Hybrid quantum--classical architecture studied in this work. Classical and quantum encoders produce complementary latent representations, which are merged via fusion strategies (early, late, or cross-attention) before classification.
  • Figure 2: Model comparison: (a) pure quantum model using variational circuits only; (b) original hybrid QML approach where a classical MLP is trained on quantum measurement outputs; (c) our proposed multimodal hybrid QML model using cross-attention fusion between classical and quantum representations.
  • Figure 3: Accuracy comparison across datasets (mean over 5-fold cross-validation). We report Classical, our best-performing Fusion variant per dataset (“Best Fusion”), and the Residual-6q-Deep baseline. Error bars denote the standard deviation across folds where available (Classical and Best Fusion); Residual-6q-Deep reports mean accuracy only. The y-axis is truncated to the 70--100% range to make high-accuracy differences visible; the Quantum-only baseline is omitted from this plot to avoid compressing the scale.