MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202541005709 A) filed by Indian Institute Of Science; and Accenture Global Solutions Limited on February 07, 2025, for Training Quantum Boltzmann Machines (qbm) Containing Hidden Variables.
Inventors include Adit Vishnu; Abhay Shastry; Chiranjib Bhattacharyya; Srikumar Karaikudi Subramanian; and Syama Sundar Santosh Kumar Gopasana.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present disclosure relates to method and system for training a Quantum Boltzmann Machine (QBM) (300) comprising visible qubits (302a–302n) and hidden qubits (306a–306n). The method comprises (a) initializing (802) model parameters of the QBM (300). Further, the method comprises (b) selecting (804) one training technique for the initialized QBM (300) from (i) an exact gradient technique (500) or (ii) a density operator expectation maximization (DO EM) technique (600A, 600B). The method further comprises (c) computing (806) an update expression for the model parameters based on the selected training technique. The method further comprises (d) updating (808) the model parameters using the update expression and (e) training (810) the QBM (300) by iteratively repeating steps (c) and (d) until a predefined convergence criterion is satisfied.
Disclaimer: Curated by HT Syndication.
