In advanced semiconductor process technology, lithography hotspot detection has become an essential task in design for manufacturability. The ability to detect and repair lithography hotspots that can affect printability is critical to improving yield and productivity. Machine learning technology has become a powerful tool in a variety of applications, from finance to manufacturing and computer vision. The use of quantum systems to process classical data using machine learning algorithms has created an emerging field of research, namely quantum machine learning (QML). We explore the possibility of converting a novel machine learning model to a hybrid quantum-classical machine learning that benefits from using variational quantum layers. We show that this hybrid model can perform similar to the classical approach. In addition, we explore parametrized quantum circuits (PQC) with different expressibility and entangling capacities. Then we compare their training performance to quantify the expected benefits. These results can be used to build a future roadmap to develop circuit-based hybrid quantum-classical machine learning for lithography hotspot detection.
In advanced semiconductor process technology, lithography hotspot detection has become an essential task in design for manufacturability. The ability to detect and repair lithography hotspots that can affect printability is critical to improving yield and productivity. Machine learning technology has become a powerful tool in a variety of applications, from finance to manufacturing and computer vision. The use of quantum systems to process classical data using machine learning algorithms has created an emerging field of research, namely quantum machine learning (QML). We explore the possibility of converting a novel machine learning model to a hybrid quantum-classical machine learning that benefits from using variational quantum layers. We show that this hybrid model can perform similar to the classical approach. In addition, we explore parametrized quantum circuits (PQC) with different expressibility and entangling capacities. Then we compare their training performance to quantify the expected benefits. These results can be used to build a future roadmap to develop circuit-based hybrid quantum-classical machine learning for lithography hotspot detection.
With the rapid development of artificial intelligence and autonomous driving technology, the demand for semiconductors is projected to rise substantially. However, the massive expansion of semiconductor manufacturing and the development of new technology will bring many defect wafers. If these defect wafers have not been correctly inspected, the ineffective semiconductor processing on these defect wafers will cause additional impact to our environment, such as excessive carbon dioxide emission and energy consumption. In this paper, we utilize the information processing advantages of quantum computing to promote the defect learning defect review (DLDR). We propose a classical-quantum hybrid algorithm for deep learning on near-term quantum processors. By tuning parameters implemented on it, quantum circuit driven by our framework learns a given DLDR task, include of wafer defect map classification, defect pattern classification, and hotspot detection. In addition, we explore parametrized quantum circuits with different expressibility and entangling capacities. These results can be used to build a future roadmap to develop circuit-based quantum deep learning for semiconductor defect detection.
The geographical dispersion of manufacturing units introduces challenges related to latency, network connectivity, data synchronization, communication overhead, scalability, data security, and resource allocation. Manufacturing data often includes sensitive information related to proprietary processes, designs, and technologies. Companies are hesitant to share this data such as lithography hotspot data (LHD) directly due to concerns about privacy and the risk of intellectual property theft. Federated learning provides a solution to address these challenges by allowing companies to collaboratively train models without sharing raw data. It allows companies to adhere to privacy laws while benefiting from shared insights. In recent years, the convergence of quantum machine learning has demonstrated transformative potential by leveraging quantum principles to address complex problems across various industries. In this work, a concept of quantum federated learning consisting of variational quantum circuit is proposed to address the lithography hotspot sharing challenges by providing a privacy-preserving, centralized, and quantum-enhanced framework. It combines the collaborative advantages of federated learning with the unique capabilities of quantum computing, offering a solution that is secure, scalable, and adaptable to the complexities of modern manufacturing environments. The performance evaluations involve scalability and robustness, particularly in the context of label-flipping attacks, which is crucial for validating, optimizing, and enhancing the quantum federated learning framework's applicability to real-world challenges in the manufacturing sector.
The emergence of deep vision models such as convolutional neural networks and vision transformers has revolutionized computer vision, enabling significant advancements in image classification, object detection, and segmentation. In parallel, the rapid development of quantum computing has spurred interest in quantum machine learning (QML), which integrates the strengths of quantum computation with the representational power of deep learning. In QML, parameterized quantum circuits offer the potential to capture complex image features, define complex decision boundaries, and provide other computational advantages. This paper investigates hybrid quantum-classical vision architectures, with a focus on hybrid quantum-classical convolutional neural networks and hybrid quantum-classical vision transformers. These hybrid models explore both quantum pre-processing and post-processing of data, respectively, where quantum circuits are strategically integrated into the data pipeline to enhance model performance. Our results suggest that these hybrid models can enhance accuracy and computational efficiency in vision-related tasks, even with the constraints of current noisy intermediate-scale quantum devices.
Machine Learning (ML) has been widely applied across numerous domains due to its ability to automatically identify informative patterns from data for various tasks. The availability of large-scale data and advanced computational power enables the development of sophisticated models and training strategies, leading to state-of-the-art performance, but it also introduces substantial challenges. Quantum Computing (QC), which exploits quantum mechanisms for computation, has attracted growing attention and significant global investment as it may address these challenges. Consequently, Quantum Machine Learning (QML), the integration of these two fields, has received increasing interest, with a notable rise in related studies in recent years. We are motivated to review these existing contributions regarding quantum circuit-based learning models for classical data analysis and highlight the identified potentials and challenges of this technique. Specifically, we focus not only on QML models, both kernel-based and neural network-based, but also on recent explorations of their integration with classical machine learning layers within hybrid frameworks. Moreover, we examine both theoretical analysis and empirical findings to better understand their capabilities, and we also discuss the efforts on noise-resilient and hardware-efficient QML that could enhance its practicality under current hardware limitations. In addition, we cover several emerging paradigms for advanced quantum circuit design and highlight the adaptability of QML across representative application domains. This study aims to provide an overview of the contributions made to bridge quantum computing and machine learning, offering insights and guidance to support its future development and pave the way for broader adoption in the coming years.
Due to the widespread presence of manufacturing units, real-time data collection becomes a challenging task, especially as the scale of operations continues to expand. Moreover, manufacturing industries often deal with sensitive and proprietary data and are reluctant to share data directly in areas like lithography hotspot detection (LHD) due to concerns about IP protection and regulatory compliance. Moreover, human error can lead to incorrect labeling of parts or products. Detecting hotspots in lithography is crucial for ensuring the product quality and reliability of the manufacturing process. In today's context, Federated Learning (FL) stands out as a crucial remedy, facilitating the rapid advancement of distributed machine learning while effectively managing critical concerns regarding data privacy and governance. Moreover, FL systems must be robust against not only malicious label flipping attacks but also inadvertent human errors. In recent years, the convergence of quantum computing and machine learning represents a groundbreaking interdisciplinary approach with immense potential for revolutionizing various industries ranging from healthcare to manufacturing. In this work, we proposed a Quantum Federated Learning (QFL) framework for lithography hotspot detection and offer a unique solution to address the challenges mentioned above effectively. To demonstrate the generalization ability and robustness, we performed extensive experiments with label-flipping attacks in the proposed framework. Based on our results, the global quantum model maintained a high classification accuracy and best performance even when faced with label-flipping attacks from a specific number of clients in federated learning settings, while keeping data private across manufacturing units. Moreover, a defense strategy based on clustering is proposed to distinguish the good and bad model updates from the participants in a quantum federated framework.
The entwining of quantum computing with classical machine learning has proven to be one of the most promising frontiers of artificial intelligence, as it possesses the potential to speed up intricate computations in ways that classical systems will never be able to achieve. This study delves into the theoretical and practical implications of hybrid quantum-classical approaches in machine learning, while also examining their advantages, limitations, and applicability in different domains. Utilizing quantum entanglement and parallelism, hybrid solutions also have the potential to process high-dimensional data, optimize complex functions, and facilitate pattern recognition tasks more efficiently. However, noise, decoherence, and limited qubit connectivity on current hardware pose considerable barriers to practical implementations. The authors discuss hybrid architectures like quantum-enhanced optimizers and variational quantum circuits which include classical pre- and post-processing to alleviate quantum limitations. Moreover, a comparative study is performed to show the effectiveness of hybrid models comparing with classical machine learning state-of-the-art in natural language processing, computer vision and financial modeling. Results indicate that although hybrid quantum–classical approaches provide significant gains in optimization and computational efficiency; these benefits remain limited by hardware limitations and the scalability of quantum algorithms. With these considerations in mind, the paper offers guidance for targeting future research directions in quantum machine learning, including advances in quantum error correction, quantum hardware scalability, and new algorithmic approaches. Additional advancements needed and notably, the study focused on a smaller representation of data Since quantum computing is relatively new and researchers are still exploring its application, it is still early days in terms of machine learning models that are solely, intrinsically quantum.