Variable Quantum Noise Training for Hybrid Quantum-Classical Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing hybrid quantum-classical machine learning models suffer from overfitting and convergence in local minima, reducing prediction accuracy and generalization capabilities.

Innovation Solution

Introduce a variable quantum noise source during training to vary the noise level in variational quantum circuits, allowing the model to improve prediction accuracy by reducing overfitting and enhancing generalization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a hybrid quantum-classical machine learning model is trained using variational quantum circuits, then the model can leverage quantum mechanical properties to solve intractable problems, but the model suffers from overfitting and convergence in local minima which reduces prediction accuracy

Engineering Contradiction:
Improveprediction accuracyVSAvoidgeneralization capability
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent applies parameter changes by systematically varying training parameters including noise levels, learning rates, and circuit depth during the training process. This allows the model to escape local minima and improve generalization by exploring different regions of the parameter space, directly addressing the contradiction between achieving high accuracy and maintaining generalization capability

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent implements dynamics by making the training process adaptive through dynamic adjustment of hyperparameters and training conditions. The system dynamically modifies training strategies based on performance metrics, enabling the model to transition from exploitation to exploration as needed, thereby resolving the tension between convergence and generalization

Inventive Principle:
Principle #15Dynamics

2Quantity of substance

If quantum gates are used to create entangled states for quantum advantage, then access to large internal state space is achieved, but the states are inherently volatile and subject to decoherence which limits the number of controllable systems and operations

Engineering Contradiction:
Improveinternal state spaceVSAvoidstate stability
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent converts the harmful effect of decoherence and noise into a beneficial training mechanism. By intentionally introducing controlled noise during training, the system prepares the quantum model to be robust against environmental interference, transforming the vulnerability of quantum states into a strength for building fault-tolerant quantum machine learning systems

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

Solution Approach 2:

The patent applies beforehand cushioning by pre-training the quantum model with artificial noise and decoherence effects. This prepares the system in advance to handle the inherent volatility of quantum states, creating a buffer against the unavoidable decoherence that occurs during actual quantum operations

Inventive Principle:
Principle #11Beforehand cushioning (Prior cushioning)

Data Source

PatentEP4625270A1A method and system for training hybrid quantum-classical machine learning models
Publication Date: 2025.10.01 TERRA QUANTUM AG
  • EP4625270A1 patent drawingFigure 1
  • EP4625270A1 patent drawingFigure 2
  • EP4625270A1 patent drawingFigure 3

AI summary

A computer-implemented method for training a hybrid quantum-classical machine learning model including a variational quantum circuit to approximate a given labeling function, the method comprising providing a variable quantum noise source in the variational quantum circuit, training the hybrid quantum-classical machine learning model based on a variation of variational parameters of the variational quantum circuit to approximate the given labeling function with the variable quantum noise source introducing a non-zero training noise level in the variational quantum circuit, and providing the hybrid quantum-classical machine learning model trained with the training noise level as a final trained hybrid quantum-classical machine learning model with the variable quantum noise source configured to introduce a noise level, which is different from the training noise level.