Bridging Quantum Computing Paradigms toward Semiconductor Yield: A Controlled CV-versus-DV Comparison on Wafer-Map Defect Classification
arXiv:2607.00961v1 Announce Type: cross Abstract: Realizing quantum neural networks (QNNs) in industry requires knowing which quantum computing paradigm suits which task. Motivated by AI accelerators and high-bandwidth memory, where die stacking makes wafer-level defect screening central to yield, we study WM-811K wafer-map defect classification (eight classes), comparing the dominant paradigms, continuous-variable (CV) and discrete-variable (DV), under controlled conditions. To isolate the quantum circuit as the sole variable, a shared convolutional backbone (~4.3M parameters) feeds interchan
The increasing complexity of AI accelerators and high-bandwidth memory necessitates advanced defect screening, and quantum computing is maturing to a point where practical applications are being explored.
Efficient wafer-map defect classification is crucial for maximizing semiconductor yield, directly impacting the cost and availability of advanced computing components vital for AI and other high-tech sectors.
This research provides a controlled comparison of quantum computing paradigms for a critical industrial problem, offering insights into which quantum approach (CV or DV) is more suitable for specific manufacturing challenges.
- · Quantum computing companies
- · Semiconductor manufacturers
- · AI hardware developers
- · Quantum algorithm researchers
- · Traditional defect classification methods
Improved semiconductor manufacturing yields due to more effective defect detection and classification.
Accelerated development and adoption of quantum computing solutions for industrial applications, particularly in manufacturing.
Potentially lower costs and increased availability of advanced AI chips and quantum hardware, driving further innovation in these fields.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG