Curator's Take
AI Commentary
This article shows that a ten‑class MNIST classifier can run end‑to‑end on IBM’s 127‑qubit Eagle processor by offloading all training to classical simulation and using a new “quantum multi‑programming” scheme to parallelise inference, thereby sidestepping the prohibitive cost of on‑hardware gradient calculations. By demonstrating that fine‑tuning on NISQ hardware offers no measurable accuracy boost, the work establishes a realistic workflow—train classically, infer quantumly—that could be adopted for other near‑term applications where quantum circuits provide unique feature maps rather than raw performance gains. While the quantum module does not yet surpass a matched classical network per parameter, the study validates multi‑class image classification on large NISQ devices and paves the way for more scalable quantum inference pipelines.
— Mark Eatherly
Summary
Quantum machine learning on real noisy intermediate-scale quantum (NISQ) hardware has remained largely confined to binary or few-class tasks, limited by the cost of on-hardware training and the underuse of large devices at inference. We present a unified framework that classifies ten-class MNIST end-to-end on a $127$-qubit IBM Eagle processor, with three central contributions. First, a two-phase protocol decouples a gradient-based classical optimization of the encoder and readout from a gradient-free optimization of the quantum parameters, removing the parameter-shift gradient cost that makes on-hardware training impractical. Second, we introduce Quantum Multi-Programming to a trained quantum classifier for the first time, packing multiple circuit copies onto one device to deliver parallel inference at no mean-accuracy cost while cutting quantum-processor job submissions proportionally. Third, a controlled comparison shows that on-hardware fine-tuning yields no measurable accuracy gain, motivating a practical NISQ workflow: train on a classical simulator and reserve the hardware for inference only. Benchmarked against a matched-capacity classical network, the quantum module shows no per-parameter accuracy advantage at this scale; we therefore frame the work as a feasibility-and-workflow demonstration for multi-class quantum image classification on current hardware.