System and method for processing language model outputs using quantum computing

Abstract

A system for data processing that includes receiving data from a classical computing environment, encoding the received data into a format compatible with a quantum computing environment which process

Description

(12) UK Patent Application (19) GB (11) 2700830 (13) A (43) Date of A Publication 18.03.2026 (21) Application No: GB2509761.9 (22) Date of Filing: 19.06.2025 (30) Priority Data: (31) 10202401825 (32) 21.06.2024 (33) SG (71) Applicant(s): Standard Chartered Bank, Singapore Branch 8 Marina Boulevard, #27-01 Marina Bay Financial Centre, Singapore, Singapore, 018981 (72) Inventor(s): Tom Pfaff, Craig Turrell (74) Agent and / or Address for Service: Marks & Clerk LLP 40 Torphichen Street, Edinburgh, EH3 8JB (51) INT CL: G06N10 / 80 (2022.01) G06F16 / 21 (2019.01) (56) Documents Cited: None (58) Field of Search: INT CL: G06N Other: No search performed 17(5)(b) (54) Title of the Invention: System and method for processing language model outputs using quantum computing Abstract Title: Structured dataset compatible with a metadata standard (57) A system for data processing that includes receiving data from a classical computing environment, encoding the received data into a format compatible with a quantum computing environment which processes the encoded data, transferring the encoded data to the quantum computing environment and transforming the processed data into a structured dataset that is compatible with a metadata standard. The structured dataset compatible with the metadata standard may correspond to an ISO 11179 compliant JSON, JavaScript Object Notation, representation. In this way, better processing of large amounts of unstructured textual data for large language models may be provided.[fig 1] Abstract: A data processing system includes: receiving data from a classical computing environment; encoding the received data into a format compatible with a quantum computing environment; processing the encoded data; transmitting the encoded data to the quantum computing environment; and converting the processed data into a structured dataset compatible with a meta-data standard. This structured dataset compatible with the meta-data standard can correspond to a JSON (JavaScript Object Notation) representation conforming to the ISO 11179 standard. This approach allows for better handling of large amounts of unstructured text data from large language models.

Claims

1. A system for data processing, comprising:a classical data interface configured to receive data from a classical computing environment;a data encoding module configured to encode the received data into a format compatible with a quantum computing environment;a quantum processing module within the quantum computing environment configured to process the encoded data;a communication interface configured to transfer the encoded data to the quantum computing environment and retrieve the processed data from the quantum computing environment; anda data decoding module configured to decode the processed data into a format compatible with the classical computing environmentwherein the data encoding module is further configured to transform the received data into a structured dataset that is compatible with a metadata standard.

2. The system of claim 1, wherein the structured dataset that is compatible with a metadata standard corresponds to an ISO 11179 compliant JSON representation.

3. The system of any one of the preceding claims, wherein the quantum processing module is configured to use at least one quantum algorithm selected from the group comprising of quantum support vector machines (QSVMs) and quantum neural networks (QNNs).

4. The system of any one of the preceding claims, further comprising a task scheduler for allocating tasks between the classical computing environment and quantum computing environment based on at least one of the following criteria: complexity of the data, suitability of processing requirements for quantum or classical computation, and available computational capacity in both computing environments.

5. The system of any preceding claim, wherein the data decoding module is configured to integrate the decoded processed data into downstream applications in the classical computing environment.

6. The system of any preceding claim, wherein the quantum processing module is configured to apply a plurality of task scheduling algorithms within the quantum computing environment based on characteristics of quantum processors.

7. The system of any preceding claim, wherein the encoded data transferred to the quantum computing environment and the processed data decoded for the classical computing environment are both in a structured data format that is compatible with a metadata standard.

8. The system of any preceding claim, wherein the system is configured to be implemented as part of a hybrid quantum-classical computing system that supports a hardware-agnostic abstraction layer which allows the computing system to interact with a plurality of quantum processors.

9. A method for transforming language model outputs, comprising:parsing an output dataset corresponding to one or more language models to extract a dataset;processing the extracted dataset to generate a refined dataset;encoding the refined dataset into a structured dataset including metadata parameters, wherein the structured dataset is configured for quantum processing; andevaluating the structured dataset for compatibility with a plurality of quantum computing systems.

10. The method of claim 9, wherein the metadata parameters comprise at least one of qubit requirements, quantum gate operations, and connectivity between qubits and a quantum processing unit.

11. The method of claim 9 or 10, wherein the structured dataset configured for quantum processing corresponds to an international protocol specific to quantum computing, high-level petri nets, or information technology.

12. The method of any one of claims 9 to 11, wherein the structured dataset is ISO 11179 compliant for providing compatibility with a plurality of quantum computing systems.

13. A method for processing language model outputs, comprising:analysing output data from one or more language models to extract a dataset;preparing the extracted dataset for compatibility with a standardized data format and quantum computing;encoding the prepared dataset into a structured dataset, wherein the structured dataset includes metadata for quantum computing operations; andevaluating the structured dataset for compatibility with a metadata standard.

14. A system for quantum natural language processing, comprising:a quantum computing device configured to execute quantum circuits for processing quantum states;a memory storing instructions executable by a processor to:encode input data into the quantum states using quantum circuits executed by the quantum computing device;apply one or more quantum operations via the quantum circuits to the encoded quantum states for processing at least one natural language task selected from the group comprising of semantic analysis, sentiment classification, named entity recognition, and text summarization;measure the encoded quantum states after the application of the one or more quantum operations to determine a result of the processing of the at least one natural language task; andoutput the determined result in a structured dataset format suitable for classical data processing.

15. The system of claim 14, wherein the one or more quantum operations include a Quantum Latent Semantic Analysis (QLSA) to identify semantic relationships within the text data.

16. The system of claim 14 or 15, wherein the quantum operations include at least one of a Quantum Support Vector Machine (QSVM), a Quantum Generative Adversarial Network (QGAN), or a Quantum Variational Autoencoder (QVAE).

17. The system of any one of claims 14-16, wherein the quantum operations include at least one of quantum entanglement and superposition principles to enhance processing speed for the at least one natural language task.

18. The system of any one of claims 14-17, wherein the structured dataset format is ISO11179 compliant for providing compatibility with a plurality of quantum computing systems.

19. The system of any one of claims 14-18, wherein the input data is provided as a structured dataset that is compatible with a metadata standard prior to encoding of the input data.

20. The system of any one of claims 14-19, wherein the quantum operations include a quantum text summarization algorithm configured to generate summaries of text by using at least one of quantum phase estimation and quantum amplitude amplification.

21. The system of any one of claims 14-20, wherein the structured dataset format includes metadata corresponding to at least one of a natural language task, a quantum operation applied, a device parameter, and a number of quantum measurements.

22. The system of any one of claims 14-21, wherein the quantum operations include the use of Hadamard and controlled gates within the quantum circuits to encode a plurality of sentiment features into the quantum states.

Images