Tailor Made Embeddings for Quantum Machine Learning
This work addresses the problem of efficiently embedding classical data into quantum states for quantum machine learning, offering a practical method that is robust to noise and requires only polynomial measurements for reconstruction.
The paper introduces a variational autoencoder framework for quantum machine learning that learns task-specific quantum embeddings of classical data, compressing high-dimensional datasets like ImageNet into a 13-qubit representation. On MNIST (3 vs 5), it achieves 98.5% validation accuracy, within 1.2 percentage points of a classical baseline and over 30 points above naive amplitude embedding.
Autoencoders transformed classical machine learning by solving the curse of dimensionality, enabling principled weight initialization and learning compact, structured representations. In this work, we extend this paradigm to quantum machine learning by introducing a variational autoencoder framework that learns task-specific quantum embeddings of classical data. We demonstrate that high-dimensional datasets, including ImageNet, can be compressed into a 13-qubit quantum representation while remaining reconstructable through a learned decoder. On MNIST (3 vs 5), our approach achieves 98.5% validation accuracy using a circuit-centric quantum classifier, within 1.2 percentage points of a classical neural network baseline (99.7%) and more than 30 percentage points above a naive amplitude-embedding approach. Unlike amplitude embeddings, which require full quantum state tomography for recovery, or angle embeddings, which generally rely on circuit inversion under restrictive assumptions, the proposed framework reconstructs the original data from only a polynomial number of measurements. The framework was further validated on IBM quantum hardware, confirming that the learned embeddings remain stable and reconstructable under real device noise.