Quantum ML Feature Grouping for Limited-Qubit Inputs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current quantum computers are limited by their small scale and inability to handle a large number of inputs, which restricts their application in machine learning tasks requiring thousands or tens of thousands of features.
Innovation Solution
A divide-and-conquer approach is employed, grouping features into subsets that match the number of inputs of the quantum computer, using multiple instances of quantum computer-based machine learning models to identify relevant feature groups and individual features iteratively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If quantum computers are used for machine learning, then computational speed is improved, but the number of inputs is limited
Solution Approach 1:
The patent divides the large set of features into multiple smaller feature groups, where each group can be processed by the quantum computer's limited number of qubits. This segmentation allows the system to handle thousands of features by processing them in manageable chunks through multiple quantum computer instances working in parallel or sequentially.
Solution Approach 2:
The patent transitions from processing individual features to processing feature groups as a higher-level abstraction. By grouping features and processing them collectively, the system effectively increases the input capacity beyond the physical limit of individual qubits, solving the contradiction between computational speed and input quantity.
2Quantity of substance
If the number of qubits is increased to handle more inputs, then the number of inputs is improved, but device complexity increases
Solution Approach 1:
The patent creates a universal quantum computer-based machine learning system that can handle variable numbers of inputs through feature grouping. Instead of building different quantum systems for different input sizes, a single framework using multiple instances of the same quantum computer architecture can adapt to various problem sizes by adjusting the number and size of feature groups.
Solution Approach 2:
The patent uses multiple copies (instances) of the quantum computer model, each processing a specific feature group. This copying approach allows the system to scale input capacity by adding more processing instances rather than increasing the complexity of individual quantum computers, maintaining manageable device complexity while handling large numbers of inputs.
Data Source
AI summary
Quantum computers with a limited number of input qubits are used to perform machine learning processes having a far greater number of trainable features. A list of features of a field are divided into a plurality of feature groups. Each of the feature groups includes a respective group of some, but not all, of the features. A first machine learning process is performed to train a first instance of a quantum computer model, where the feature groups are used as inputs. Based on the first machine learning process being performed, a subset of the feature groups is selected for a second machine learning process. Thereafter, the second machine learning process is performed to train one or more second instances of the quantum computer model. The individual features of the selected subset of the feature groups are used as inputs for the second instances of the quantum computer model.