Evolutionary quantum neural network architecture search method based on quantum simulator

Abstract

The invention discloses an evolutionary quantum neural network architecture searching method based on a quantum simulator. The evolutionary quantum neural network architecture searching method mainly solves the problems that a quantum neural network model designed in the prior art is low in precision and high in complexity. According to the implementation scheme, quantization coding is carried out on image data; designing a basic framework of the quantum neural network; using the image data after quantum coding to search optimal structure parameters under the basic framework of the quantum neural network by adopting a quantum evolutionary algorithm, in the quantum evolutionary algorithm, coding the quantum neural network into quantum chromosomes, and using quantum observation, quantum revolving door updating and full interference crossover operation to search the optimal structure parameters; and constructing an optimal quantum neural network based on the optimal structure parameters. The quantum neural network obtained by searching has higher model precision and lower complexity, can be deployed on a quantum simulator or a real quantum system, makes full use of the parallel advantage of quantum calculation, improves the reasoning speed of the model, and can be used for image classification.

Description

technical field [0001] The invention belongs to the technical field of quantum computing, in particular to an evolutionary quantum neural network architecture search method, which can be used for image classification. Background technique [0002] A quantum simulator is a simulation of a quantum computer implemented on a classical computer. In a quantum simulator, qubits, quantum circuits, and quantum environments will all simulate a real quantum computer. In order to make it easier for researchers to study and design quantum algorithms, many companies and institutions have designed and developed a large number of quantum simulators, and Cirq developed by Google is one of them. Cirq provides users with a large number of quantum logic gate interfaces and methods for manipulating qubits, allowing users to precisely control quantum circuits and build quantum neural networks. TensorFlow Quantum (TFQ) is an open-source software framework developed by Google researchers for the ...

AI Technical Summary

Problems solved by technology

Reinforcement learning-based and evolution-based methods can search for more efficient network architectures that are different from artificial design, but at the same time have higher time costs

The gradient-based method uses a gradient descent optimization strategy to reduce the search time overhead, but it is easier to fall into a local optimum

For quantum neural network neural architecture search, traditional neural network architecture search methods cannot fully utilize the parallel characteristics of quantum, and cannot be fully run in quantum simulators or quantum circuits, so they cannot be fully transplanted to quantum computers. Secondly, based on evolutionary neural network The population diversity and convergence of the architecture search method cannot be guaranteed in the later stages of the population iteration, and the local search ability is insufficient

In addition, because the quantum neural network is still in the exploratory stage and does not have a unified and recognized structure, there is no architecture search method and related theories for quantum neural networks.

Images