Quantum Machine Learning Classifier With Ising Ground-State Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current quantum computers face challenges in processing classical data due to limited capacity and exponential quantum state spaces, leading to model overfitting and difficulties in encoding data effectively.
Innovation Solution
A method involving non-unitary measurements on an auxiliary system to project the encoded model into the ground state, using quantum processing units or their emulations, and extracting information as observables for predictive modeling, with techniques like Ising Hamiltonian encoding and quantum circuit execution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If data is encoded using native control-Z operations and virtual Z rotations on quantum processing hardware, then the classical problem can be expressed on a quantum computer with hardware efficiency, but the exponential quantum state space causes model overfitting
Solution Approach 1:
The patent extracts only the necessary quantum features from the full quantum state space by performing measurements on a subset of qubits. This selective extraction prevents the model from fitting to the entire exponential state space, thereby avoiding overfitting while maintaining hardware-efficient encoding through control-Z operations and virtual Z rotations.
Solution Approach 2:
The patent applies partial measurement action by measuring only certain qubits rather than the entire quantum state. This partial action is sufficient to capture the essential classification information while avoiding the harmful effect of capturing noise from the full exponential state space, thus preventing overfitting.
2Loss of information
If the full quantum state space is utilized for encoding classical data, then comprehensive information representation is achieved, but the limited processing capacity of current and near-term quantum computers becomes insufficient
Solution Approach 1:
The patent extracts the essential information needed for classification by measuring only a subset of qubits rather than the full quantum state. This extraction maintains sufficient information representation for the classification task while being compatible with the limited processing capacity of current quantum computers.
Solution Approach 2:
The patent uses partial measurement action on selected qubits, which provides sufficient information for effective classification without requiring the full processing capacity needed to handle the entire exponential quantum state space.
3Reliability
If non-unitary measurements are performed on an auxiliary system to project the encoded model into the ground state, then model overfitting is prevented and classification accuracy is improved, but additional measurement operations increase system complexity
Solution Approach 1:
The patent uses an auxiliary system of qubits as an intermediary to perform the classification measurement. The auxiliary qubits interact with the encoded data qubits through controlled operations, enabling the projection into the ground state and extraction of classification information without requiring direct complex measurements on the full system.
Solution Approach 2:
The patent segments the quantum system into data qubits for encoding and auxiliary qubits for measurement. This segmentation allows the classification function to be separated from the data storage function, simplifying the measurement process while maintaining classification accuracy and preventing overfitting.
Data Source
AI summary
Methods and systems for training and using a binary classifier implemented using quantum computing techniques are disclosed. The described approach involves deriving, from an input data set, a plurality of training samples, each training sample comprising a data vector having a plurality of features and a class label. Each data vector is processed using a quantum classification process including: encoding the data vector as an Ising Hamiltonian; implementing the Ising Hamiltonian on a set of real or virtual qubits of a quantum processing unit or an emulation thereof to form a quantum system representing the data vector; executing operations on the (emulation of the) quantum processing unit to prepare the ground state of the quantum system; determining one or more properties of the ground state; and identifying one of a set of possible ground states corresponding to the data vector based on the one or more properties. The system then determines, based on the identified ground states and class labels for the training samples, a mapping that maps ground states to class labels. The mapping is stored and used for classifying further data samples.