Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2511.00406

Quantum Machine Unlearning: Foundations, Mechanisms, and Taxonomy

Thanveer Shaik, Xiaohui Tao, Haoran Xie

Published Nov 1, 2025• 4 citations
arXivHTMLPDF

Abstract

Quantum Machine Unlearning has emerged as a foundational challenge at the intersection of quantum information theory privacypreserving computation and trustworthy artificial intelligence This paper advances QMU by establishing a formal framework that unifies physical constraints algorithmic mechanisms and ethical governance within a verifiable paradigm We define forgetting as a contraction of distinguishability between pre and postunlearning models under completely positive trace-preserving dynamics grounding data removal in the physics of quantum irreversibility Building on this foundation we present a fiveaxis taxonomy spanning scope guarantees mechanisms system context and hardware realization linking theoretical constructs to implementable strategies Within this structure we incorporate influence and quantum Fisher information weighted updates parameter reinitialization and kernel alignment as practical mechanisms compatible with noisy intermediatescale quantum NISQ devices The framework extends naturally to federated and privacyaware settings via quantum differential privacy homomorphic encryption and verifiable delegation enabling scalable auditable deletion across distributed quantum systems Beyond technical design we outline a forwardlooking research roadmap emphasizing formal proofs of forgetting scalable and secure architectures postunlearning interpretability and ethically auditable governance Together these contributions elevate QMU from a conceptual notion to a rigorously defined and ethically aligned discipline bridging physical feasibility algorithmic verifiability and societal accountability in the emerging era of quantum intelligence.

DOI
10.48550/arxiv.2511.00406

Related papers

Post-Quantum AI: Building Secure Machine Learning Systems in the Quantum Era

Amit Taneja

This book explores the intersection of artificial intelligence (AI) and quantum computing, focusing on the urgent need to secure machine learning systems in the face of emerging quantum threats. As quantum computers advance, they expose vulnerabilities in classical cryptographic methods, potentially undermining data integrity, privacy, and trust in AI-driven applications. To address these challenges, this study introduces the concept of post-quantum AI—a framework for integrating quantum-resistant cryptographic algorithms, Anomaly detection mechanisms, and Resilient machine learning architectures. This book makes three core contributions: it motivates a quantum-era threat model for machine learning (ii) it maps a migration path to standardised post-quantum cryptography Crypto-agile architectures (iii) it presents a defence-in-depth blueprint across the data → training → inference lifecycle that integrates privacy-preserving learning Governance. This work explicitly highlights key contributions, including proposed frameworks, algorithms, and case studies. Future research directions are also outlined to guide continued exploration in this emergent field. The final candidate algorithms from the NIST PQC standardisation process (NIST, 2022–2023) further strengthen this discussion. Key themes include the foundations of quantum mechanics relevant to computation, the fundamental differences between classical and quantum computing, and the transformative potential of quantum algorithms for optimisation, pattern recognition, and predictive analytics. The book highlights case studies spanning drug discovery, finance, mobile networks, and supply chain optimisation, illustrating how quantum-enhanced AI can revolutionise industry while simultaneously raising complex security and ethical concerns. A central focus is the development and deployment of post-quantum cryptography (PQC)), such as lattice-based and hash-based algorithms, to safeguard AI models against adversarial and quantum-accelerated attacks. The discussion extends to adversarial machine learning, explainable AI (XAI), and hybrid classical–quantum systems as strategies for strengthening resilience. The ethical, legal, and regulatory dimensions of post-quantum AI are also examined, emphasising fairness, transparency, accountability, and international cooperation. By combining technical innovations with responsible governance, the book advocates for building trustworthy AI systems that remain robust in the quantum era. Future work includes post-quantum cryptography (PQC) performance benchmarking in ML pipelines Patterns for crypto-agile key management, Assurance methods for privacy-preserving Federated learning as standards, and Implementations that are mature.

Book Publisher International (a part of SCIENCEDOMAIN International)•Sep 27, 2025•0 citations

Trustworthy Quantum Machine Learning: A Roadmap for Reliability, Robustness, and Security in the NISQ Era

Ferhat Ozgur Catak, Jungwon Seo, Umit Cali

Quantum machine learning (QML) is a promising paradigm for tackling computational problems that challenge classical AI. Yet, the inherent probabilistic behavior of quantum mechanics, device noise in NISQ hardware, and hybrid quantum-classical execution pipelines introduce new risks that prevent reliable deployment of QML in real-world, safety-critical settings. This research offers a broad roadmap for Trustworthy Quantum Machine Learning (TQML), integrating three foundational pillars of reliability: (i) uncertainty quantification for calibrated and risk-aware decision making, (ii) adversarial robustness against classical and quantum-native threat models, and (iii) privacy preservation in distributed and delegated quantum learning scenarios. We formalize quantum-specific trust metrics grounded in quantum information theory, including a variance-based decomposition of predictive uncertainty, trace-distance-bounded robustness, and differential privacy for hybrid learning channels. To demonstrate feasibility on current NISQ devices, we validate a unified trust assessment pipeline on parameterized quantum classifiers, uncovering correlations between uncertainty and prediction risk, an asymmetry in attack vulnerability between classical and quantum state perturbations, and privacy-utility trade-offs driven by shot noise and quantum channel noise. This roadmap seeks to define trustworthiness as a first-class design objective for quantum AI.

quant-pharXiv:2511.02602•Nov 5, 2025•7 citations

From membership-privacy leakage to quantum machine unlearning

Junjian Su, Runze He, Guanghui Li +4 more

Quantum machine learning (QML) has the potential to achieve quantum advantage for specific tasks by combining quantum computation with classical machine learning (ML). In classical ML, a significant challenge is membership-privacy leakage, whereby an attacker can infer from model outputs whether specific data were used in training. When specific data are required to be withdrawn, removing their influence from the trained model becomes necessary. Machine unlearning (MU) addresses this issue by enabling the model to forget the withdrawn data, thereby preventing membership-privacy leakage. However, this leakage remains underexplored in QML. This raises two research questions: do QML models leak membership privacy about their training data, and can MU methods efficiently mitigate such leakage in QML models? We investigate these questions using two quantum neural network (QNN) architectures, a basic QNN and a hybrid QNN, evaluated in noiseless simulations and cloud quantum device demonstrations. To answer the first question, we analyze how quantum constraints shape membership-privacy leakage in QML and then formalize a realistic gray-box threat model accordingly. Based on this, we design a membership inference attack (MIA) tailored to QNN outputs, and our results provide clear evidence of membership leakage in both QNNs. To answer the second question, we propose a quantum machine unlearning (QMU) framework, comprising three MU mechanisms. Evaluations on two QNN architectures show that QMU removes the influence of the withdrawn data while preserving accuracy for retained data. A comparative analysis further characterizes the three MU mechanisms with respect to data dependence, computational cost, and robustness.

arXiv:2509.06086•Physical Review Applied•Mar 23, 2026•4 citations

Quantum AI and deep-tech innovations: A comprehensive review of next-generation intelligence

Thamizhmaran K., Vimali R.

Quantum Artificial Intelligence (Quantum AI) embodies the integration of quantum computational paradigms with artificial intelligence frameworks to address the computational intractability and scalability constraints inherent in classical AI systems. By leveraging quantum mechanical properties such as superposition, entanglement, and quantum interference, Quantum AI enables enhanced representation learning, accelerated optimization, and improved probabilistic inference in high-dimensional problem spaces. In parallel, deep-tech innovations—including advanced deep learning architectures, neuromorphic computing, and edge-based intelligent processing—are redefining the architectural foundations of next- generation intelligent systems through energy-efficient, adaptive, and real-time computation. This paper presents a technically rigorous and comprehensive review of Quantum AI and deep-tech innovations, focusing on algorithmic formulations, architectural models, and system-level integration strategies. Core methodologies such as quantum machine learning algorithms, parameterized quantum circuits, variational quantum algorithms, and hybrid quantum–classical optimization frameworks are critically analysed with respect to computational complexity, repressibility, and feasibility on noisy intermediate-scale quantum (NISQ) hardware. Additionally, the role of neuromorphic computing in enabling spiking neural architectures and event-driven processing is examined as a complementary paradigm to quantum-enhanced learning models, particularly for low-power and latency-critical applications. The paper further investigates application- driven deployments of Quantum AI across domains including medical diagnostics, cryptographic security, autonomous decision-making, and smart cyber-physical systems, highlighting performance trade-offs and implementation constraints. Key challenges such as quantum noise, decoherence, data encoding bottlenecks, limited qubit connectivity, and ethical governance are systematically discussed. Finally, future research directions are identified, emphasizing fault-tolerant quantum learning architectures, scalable hybrid integration with deep-tech systems, and co-design methodologies for realizing practical next-generation intelligence platforms.

i-manager’s Journal on Software Engineering•Jan 30, 2026•0 citations

Quantum-Enhanced Artificial Intelligence: Bridging Quantum Physics and Machine Learning for Next-Generation Computing Paradigms

Mohammad Wali Khurami, Musawer Hakimi

Quantum computing and artificial intelligence (AI) are converging into a distinct research frontier commonly referred to as quantum-enhanced artificial intelligence, or quantum machine learning (QML). This paper presents a conceptual and integrative review of how principles from quantum physics superposition, entanglement, and interference can be embedded into machine learning pipelines to reshape computational paradigms for classification, optimization, and representation learning. Using a structured narrative-review methodology, the study synthesizes theoretical foundations, algorithmic building blocks (quantum feature maps, variational quantum circuits, quantum kernel methods), and application domains spanning drug discovery, finance, materials science, and natural language processing. The review develops a hybrid quantum-classical architecture model and a complexity-comparison framework contrasting classical algorithms with their quantum counterparts, including Grover's search and Shor's factoring algorithm. Findings indicate that while theoretical speedups are well established, practical quantum advantage on noisy intermediate-scale quantum (NISQ) hardware remains constrained by decoherence, barren plateaus, and limited qubit connectivity. The paper contributes a synthesized taxonomy of quantum-enhanced AI methods and an evidence-based research agenda emphasizing error mitigation, hardware-aware ansatz design, and hybrid workload partitioning. The discussion further situates these developments within the broader trajectory of next-generation computing, arguing that near-term value will accrue primarily through hybrid quantum-classical systems rather than fully quantum pipelines. Implications for researchers, industry practitioners, and policymakers are discussed, alongside limitations inherent to a literature-synthesis approach.

Buana Information Technology and Computer Sciences (BIT and CS)•Jul 30, 2026•0 citations

Quantum Foundation Models, A New Paradigm for AI Beyond the Turing Limit

Muhammad Ahmad Shahid, Shehryar Ali, Abdur Rafay Goraya

Quantum foundation models mark a transformative convergence of quantum computing and artificial intelligence, offering a novel framework for addressing computational challenges beyond the reach of classical systems. By harnessing core quantum principles such as superposition and entanglement these models enable parallel processing at unprecedented scales, significantly enhancing the efficiency and capability of machine learning algorithms. This integration opens pathways for innovation across domains like healthcare, finance, and robotics, where classical AI often encounters limitations with high-dimensional data and complex optimization tasks. Despite their immense promise, quantum foundation models face substantial hurdles, including hardware immaturity, algorithmic bias, and unresolved ethical concerns. The evolution of this field demands robust theoretical models, responsible governance, and interdisciplinary collaboration to ensure that technological advancement proceeds with transparency, fairness, and societal benefit. As research deepens, quantum foundation models are poised to redefine AI's operational and ethical landscape, setting the stage for a new era of intelligent systems.

Annual Methodological Archive Research Review•Jul 23, 2025•1 citation
Quantum Machine Unlearning has emerged as a foundational challenge at the intersection of quantum information theory privacypreserving computation and trustworthy artificial intelligence This paper advances QMU by establishing a formal framework that unifies physical constraints algorithmic mechanisms and ethical governance within a verifiable paradigm We define forgetting as a contraction of distinguishability between pre and postunlearning models under completely positive trace-preserving dynamics grounding data removal in the physics of quantum irreversibility Building on this foundation we present a fiveaxis taxonomy spanning scope guarantees mechanisms system context and hardware realization linking theoretical constructs to implementable strategies Within this structure we incorporate influence and quantum Fisher information weighted updates parameter reinitialization and kernel alignment as practical mechanisms compatible with noisy intermediatescale quantum NISQ devices The framework extends naturally to federated and privacyaware settings via quantum differential privacy homomorphic encryption and verifiable delegation enabling scalable auditable deletion across distributed quantum systems Beyond technical design we outline a forwardlooking research roadmap emphasizing formal proofs of forgetting scalable and secure architectures postunlearning interpretability and ethically auditable governance Together these contributions elevate QMU from a conceptual notion to a rigorously defined and ethically aligned discipline bridging physical feasibility algorithmic verifiability and societal accountability in the emerging era of quantum intelligence.
This book explores the intersection of artificial intelligence (AI) and quantum computing, focusing on the urgent need to secure machine learning systems in the face of emerging quantum threats. As quantum computers advance, they expose vulnerabilities in classical cryptographic methods, potentially undermining data integrity, privacy, and trust in AI-driven applications. To address these challenges, this study introduces the concept of post-quantum AI—a framework for integrating quantum-resistant cryptographic algorithms, Anomaly detection mechanisms, and Resilient machine learning architectures. This book makes three core contributions: it motivates a quantum-era threat model for machine learning (ii) it maps a migration path to standardised post-quantum cryptography Crypto-agile architectures (iii) it presents a defence-in-depth blueprint across the data → training → inference lifecycle that integrates privacy-preserving learning Governance. This work explicitly highlights key contributions, including proposed frameworks, algorithms, and case studies. Future research directions are also outlined to guide continued exploration in this emergent field. The final candidate algorithms from the NIST PQC standardisation process (NIST, 2022–2023) further strengthen this discussion. Key themes include the foundations of quantum mechanics relevant to computation, the fundamental differences between classical and quantum computing, and the transformative potential of quantum algorithms for optimisation, pattern recognition, and predictive analytics. The book highlights case studies spanning drug discovery, finance, mobile networks, and supply chain optimisation, illustrating how quantum-enhanced AI can revolutionise industry while simultaneously raising complex security and ethical concerns. A central focus is the development and deployment of post-quantum cryptography (PQC)), such as lattice-based and hash-based algorithms, to safeguard AI models against adversarial and quantum-accelerated attacks. The discussion extends to adversarial machine learning, explainable AI (XAI), and hybrid classical–quantum systems as strategies for strengthening resilience. The ethical, legal, and regulatory dimensions of post-quantum AI are also examined, emphasising fairness, transparency, accountability, and international cooperation. By combining technical innovations with responsible governance, the book advocates for building trustworthy AI systems that remain robust in the quantum era. Future work includes post-quantum cryptography (PQC) performance benchmarking in ML pipelines Patterns for crypto-agile key management, Assurance methods for privacy-preserving Federated learning as standards, and Implementations that are mature.
Quantum machine learning (QML) is a promising paradigm for tackling computational problems that challenge classical AI. Yet, the inherent probabilistic behavior of quantum mechanics, device noise in NISQ hardware, and hybrid quantum-classical execution pipelines introduce new risks that prevent reliable deployment of QML in real-world, safety-critical settings. This research offers a broad roadmap for Trustworthy Quantum Machine Learning (TQML), integrating three foundational pillars of reliability: (i) uncertainty quantification for calibrated and risk-aware decision making, (ii) adversarial robustness against classical and quantum-native threat models, and (iii) privacy preservation in distributed and delegated quantum learning scenarios. We formalize quantum-specific trust metrics grounded in quantum information theory, including a variance-based decomposition of predictive uncertainty, trace-distance-bounded robustness, and differential privacy for hybrid learning channels. To demonstrate feasibility on current NISQ devices, we validate a unified trust assessment pipeline on parameterized quantum classifiers, uncovering correlations between uncertainty and prediction risk, an asymmetry in attack vulnerability between classical and quantum state perturbations, and privacy-utility trade-offs driven by shot noise and quantum channel noise. This roadmap seeks to define trustworthiness as a first-class design objective for quantum AI.
Quantum machine learning (QML) has the potential to achieve quantum advantage for specific tasks by combining quantum computation with classical machine learning (ML). In classical ML, a significant challenge is membership-privacy leakage, whereby an attacker can infer from model outputs whether specific data were used in training. When specific data are required to be withdrawn, removing their influence from the trained model becomes necessary. Machine unlearning (MU) addresses this issue by enabling the model to forget the withdrawn data, thereby preventing membership-privacy leakage. However, this leakage remains underexplored in QML. This raises two research questions: do QML models leak membership privacy about their training data, and can MU methods efficiently mitigate such leakage in QML models? We investigate these questions using two quantum neural network (QNN) architectures, a basic QNN and a hybrid QNN, evaluated in noiseless simulations and cloud quantum device demonstrations. To answer the first question, we analyze how quantum constraints shape membership-privacy leakage in QML and then formalize a realistic gray-box threat model accordingly. Based on this, we design a membership inference attack (MIA) tailored to QNN outputs, and our results provide clear evidence of membership leakage in both QNNs. To answer the second question, we propose a quantum machine unlearning (QMU) framework, comprising three MU mechanisms. Evaluations on two QNN architectures show that QMU removes the influence of the withdrawn data while preserving accuracy for retained data. A comparative analysis further characterizes the three MU mechanisms with respect to data dependence, computational cost, and robustness.
Quantum Artificial Intelligence (Quantum AI) embodies the integration of quantum computational paradigms with artificial intelligence frameworks to address the computational intractability and scalability constraints inherent in classical AI systems. By leveraging quantum mechanical properties such as superposition, entanglement, and quantum interference, Quantum AI enables enhanced representation learning, accelerated optimization, and improved probabilistic inference in high-dimensional problem spaces. In parallel, deep-tech innovations—including advanced deep learning architectures, neuromorphic computing, and edge-based intelligent processing—are redefining the architectural foundations of next- generation intelligent systems through energy-efficient, adaptive, and real-time computation. This paper presents a technically rigorous and comprehensive review of Quantum AI and deep-tech innovations, focusing on algorithmic formulations, architectural models, and system-level integration strategies. Core methodologies such as quantum machine learning algorithms, parameterized quantum circuits, variational quantum algorithms, and hybrid quantum–classical optimization frameworks are critically analysed with respect to computational complexity, repressibility, and feasibility on noisy intermediate-scale quantum (NISQ) hardware. Additionally, the role of neuromorphic computing in enabling spiking neural architectures and event-driven processing is examined as a complementary paradigm to quantum-enhanced learning models, particularly for low-power and latency-critical applications. The paper further investigates application- driven deployments of Quantum AI across domains including medical diagnostics, cryptographic security, autonomous decision-making, and smart cyber-physical systems, highlighting performance trade-offs and implementation constraints. Key challenges such as quantum noise, decoherence, data encoding bottlenecks, limited qubit connectivity, and ethical governance are systematically discussed. Finally, future research directions are identified, emphasizing fault-tolerant quantum learning architectures, scalable hybrid integration with deep-tech systems, and co-design methodologies for realizing practical next-generation intelligence platforms.
Quantum computing and artificial intelligence (AI) are converging into a distinct research frontier commonly referred to as quantum-enhanced artificial intelligence, or quantum machine learning (QML). This paper presents a conceptual and integrative review of how principles from quantum physics superposition, entanglement, and interference can be embedded into machine learning pipelines to reshape computational paradigms for classification, optimization, and representation learning. Using a structured narrative-review methodology, the study synthesizes theoretical foundations, algorithmic building blocks (quantum feature maps, variational quantum circuits, quantum kernel methods), and application domains spanning drug discovery, finance, materials science, and natural language processing. The review develops a hybrid quantum-classical architecture model and a complexity-comparison framework contrasting classical algorithms with their quantum counterparts, including Grover's search and Shor's factoring algorithm. Findings indicate that while theoretical speedups are well established, practical quantum advantage on noisy intermediate-scale quantum (NISQ) hardware remains constrained by decoherence, barren plateaus, and limited qubit connectivity. The paper contributes a synthesized taxonomy of quantum-enhanced AI methods and an evidence-based research agenda emphasizing error mitigation, hardware-aware ansatz design, and hybrid workload partitioning. The discussion further situates these developments within the broader trajectory of next-generation computing, arguing that near-term value will accrue primarily through hybrid quantum-classical systems rather than fully quantum pipelines. Implications for researchers, industry practitioners, and policymakers are discussed, alongside limitations inherent to a literature-synthesis approach.
Quantum foundation models mark a transformative convergence of quantum computing and artificial intelligence, offering a novel framework for addressing computational challenges beyond the reach of classical systems. By harnessing core quantum principles such as superposition and entanglement these models enable parallel processing at unprecedented scales, significantly enhancing the efficiency and capability of machine learning algorithms. This integration opens pathways for innovation across domains like healthcare, finance, and robotics, where classical AI often encounters limitations with high-dimensional data and complex optimization tasks. Despite their immense promise, quantum foundation models face substantial hurdles, including hardware immaturity, algorithmic bias, and unresolved ethical concerns. The evolution of this field demands robust theoretical models, responsible governance, and interdisciplinary collaboration to ensure that technological advancement proceeds with transparency, fairness, and societal benefit. As research deepens, quantum foundation models are poised to redefine AI's operational and ethical landscape, setting the stage for a new era of intelligent systems.