MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641107935 A) filed by Svs Group Of Institutions on September 08, 2026, for A System And Method For Hybrid Quantum-Classical Machine Learning-Based Classification.
Inventors include Dr. A. Pratapa Reddy; Dr. G. Thirupathi; Dr. G. Rajanikar; Dr. K. Yakub Reddy; Dr Anvesh Thatikonda; Dr. Yasin Ali; Mariya Nishath; Arsham Chandana; Potharaboina Vijendar; and Medipalli Ramya.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The disclosure herein pertains to a computer-implemented system and method for Hybrid Quantum-Classical Machine Learning-Based Classification that integrates quantum computing, classical artificial intelligence, explainable machine learning, and adaptive optimization for intelligent classification of structured, semi-structured, and unstructured datasets across heterogeneous application domains. The disclosed system includes a Data Acquisition Module that acquires a multitude of datasets from Electronic Health Records (EHRs), biomedical repositories, financial transaction systems, industrial automation platforms, Internet of Things (IoT) devices, distributed data sources, cloud computing infrastructures, enterprise information systems, scientific databases, cybersecurity repositories, remote sensing platforms, smart devices, and biomedical repositories. A Data Preprocessing and Feature Engineering Module ensures data validation, normalization, missing value imputation, noise removal, feature extraction, feature transformation, feature selection, categorical encoding, dimensionality reduction and feature optimization to produce standardized feature representations that are appropriate for hybrid quantum-classical learning. A Quantum Feature Encoding Module that maps optimized classical feature vectors into quantum state representations, preserving high dimensional nonlinear relationships between features, such as amplitude encoding, angle encoding, basis encoding, phase encoding, quantum embedding, data re-uploading and hybrid quantum feature encoding. To learn optimized feature representations and intelligent decision boundaries for accurate classification, A Hybrid Quantum-Classical Learning Module uses Variational Quantum Circuits (VQCs), Parameterized Quantum Circuits (PQCs), Quantum Neural Networks (QNNs), Quantum Kernel Methods (QKMs), Quantum Support Vector Machines (QSVMs), Deep Neural Networks (DNNs), Convolutional Neural Networks (CNNs), Graph Neural Networks (GNNs), Vision Transformers (ViTs), ensemble learning models, and reinforcement learning techniques. Using Gradient Descent, Adam, AdamW, RMS Prop, Bayesian optimization, Parameter shift optimization, Quantum Natural Gradient (QNG), and evolutionary optimization, the Classical Optimization Module optimizes quantum circuit parameters, classical learning parameters, objective functions, qubit allocation strategies, entanglement configurations and optimization variables to accelerate convergence, enhance the accuracy of predictions, reduce computing requirements, and optimize the use of quantum resources. An Intelligent Classification and Decision Module carries out binary classification, multi-class classification, multi-label classification and anomaly detection. Predictive Analytics, Uncertainty estimation, Probabilistic Prediction, Intelligent Decision. Support and provide a confidence score, and classification results. A Quantum The Classification Visualization and Explainability Module provides interpretable explanations of classification, feature explanation, feature importance mapping, receiver operating characteristic (ROC) curves, precision-recall analysis, confusion matrices and interactive classification analytical dashboards based on Explainable Artificial Intelligence (XAI) techniques such as SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), Integrated Gradients, attention visualization and counterfactual reasoning. An adaptive optimization scheme continuously optimizes the intelligent classification performance by evaluating the classification accuracy, precision, recall, F1-score, Area Under Curve (AUC), Matthews Correlation Coefficient (MCC), balanced accuracy, prediction reliability, optimization convergence, quantum circuit depth, qubit utilization and execution latency. An Adaptive Hybrid Learning and Model Update Module continuously updates hybrid quantum-classical learning models using newly acquired datasets, validated prediction outcomes, optimized quantum circuit parameters, benchmark datasets, feature representation updates, transfer learning, continual learning, reinforcement learning, meta-learning, and user feedback to improve future intelligent classification accuracy while preventing catastrophic forgetting. The disclosed invention further incorporates secure authentication, encrypted communication, role-based access control, federated learning, differential privacy, homomorphic encryption, secure multi-party computation (SMPC), quantum-safe cryptographic mechanisms, audit logging, and privacy-preserving collaborative learning to ensure secure deployment across distributed computing environments. The disclosed invention provides highly accurate, explainable, scalable, secure, and computationally efficient intelligent classification, optimized quantum resource utilization, adaptive model learning, transparent decision support, and seamless interoperability with cloud computing platforms, quantum computing infrastructures, enterprise information systems, healthcare platforms, financial systems, industrial automation environments, Internet of Things (IoT) ecosystems, scientific computing infrastructures, and other next-generation intelligent data-driven applications.
Disclaimer: Curated by HT Syndication.
