Abstract:This paper presents a new hybrid Quantum Machine Learning (QML) model composed of three elements: a classical computer in charge of the data preparation and interpretation; a Gate-based Quantum Computer running the Variational Quantum Algorithm (VQA) representing the Quantum Neural Network (QNN); and an adiabatic Quantum Computer where the optimization function is executed to find the best parameters for the VQA. As of the moment of this writing, the majority of QNNs are being trained using gradient-based classical optimizers having to deal with the barren-plateau effect. Some gradient-free classical approaches such as Evolutionary Algorithms have also been proposed to overcome this effect. To the knowledge of the authors, adiabatic quantum models have not been used to train VQAs. The paper compares the results of gradient-based classical algorithms against adiabatic optimizers showing the feasibility of integration for gate-based and adiabatic quantum computing models, opening the door to modern hybrid QML approaches for High Performance Computing.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to use adiabatic quantum computing (AQC) to optimize the parameter training process in variational quantum algorithms (VQAs). Specifically, the paper proposes a new hybrid quantum machine learning (QML) model, which combines the advantages of classical computers, gate - based quantum computers and adiabatic quantum computers to improve the performance of time - series prediction tasks, especially in stock price trend prediction. ### Main problem background 1. **Variational quantum algorithms (VQAs)** - VQAs are a type of quantum machine - learning model, similar to classical neural networks, which optimize parameters by adjusting the rotation angles of quantum gates. - Currently, most VQAs are trained using gradient - based classical optimizers, but this method is prone to the "barren plateau effect", that is, the gradient becomes very small in high - dimensional space, leading to training difficulties. 2. **Adiabatic quantum computing (AQC)** - AQC is a quantum computing method that finds the optimal solution to a problem by slowly changing the Hamiltonian of the system. - Unlike gradient - based methods, AQC does not rely on gradients and can thus avoid the "barren plateau effect". ### Main contributions of the paper - **Proposing a new hybrid QML model**: This model consists of three parts: - The classical computer is responsible for data preparation and interpretation. - The gate - based quantum computer runs the VQA, representing the quantum neural network (QNN). - The adiabatic quantum computer executes the optimization function to find the best parameters for the VQA. - **Comparing the effects of different optimization methods**: The paper compares the results of gradient - based classical optimization algorithms and adiabatic optimizers, demonstrating the feasibility of integrating gate - based quantum computing and adiabatic quantum computing models together, providing new possibilities for high - performance computing (HPC). ### Specific implementation - **Quantum recurrent neural network (QRNN)**: The paper describes in detail the structure of the QRNN, including input data encoding, Ansatz sub - circuits and partial measurements. - **QUBO optimization problem**: The problem of finding the optimal parameters of the QNN is transformed into an unconstrained quadratic binary optimization (QUBO) problem and solved by adiabatic quantum computing. - **Experimental results**: The paper shows the performance comparison of the QRNN optimized by adiabatic quantum, the gradient - based classical optimization method and the traditional artificial neural network (ANN) in the stock price trend prediction task. ### Conclusions - **Similar accuracy**: The experimental results show that there is no significant difference in accuracy between the QRNN optimized by adiabatic quantum and the gradient - based classical optimization method. - **Energy - efficiency advantage**: Although it has not yet been proven that quantum computers will be faster in terms of speed, according to C.H. Bennett's view, quantum computers have an advantage in terms of energy efficiency. - **Future research directions**: It is recommended to explore the introduction of imaginary - part phases or iterative approximation methods to improve the expressive ability of the model, and to verify the scalability and robustness of the model on more diverse data sets. Through these methods, the paper provides new ideas and directions for the future application of quantum computing in the field of machine learning.

Adiabatic training for Variational Quantum Algorithms

QNEAT: Natural Evolution of Variational Quantum Circuit Architecture

Learning to Learn Variational Quantum Algorithm

Training Neural Networks with Universal Adiabatic Quantum Computing

Toward Neural Network Simulation of Variational Quantum Algorithms

Variational Quantum-Neural Hybrid Eigensolver.

A semi-agnostic ansatz with variable structure for quantum machine learning

Quantum agents in the Gym: a variational quantum algorithm for deep Q-learning

Hybrid Heuristic Algorithms for Adiabatic Quantum Machine Learning Models

Learning To Optimize Quantum Neural Network Without Gradients

Neural-network-encoded variational quantum algorithms

On the relation between trainability and dequantization of variational quantum learning models

VQNet: Library for a Quantum-Classical Hybrid Neural Network

Neural network encoded variational quantum algorithms

Leveraging Pre-Trained Neural Networks to Enhance Machine Learning with Variational Quantum Circuits

Adiabatic Quantum Support Vector Machines

QUBO formulations for training machine learning models

Enhancing Variational Quantum Circuit Training: An Improved Neural Network Approach for Barren Plateau Mitigation

Optimizing Variational Quantum Algorithms with qBang: Efficiently Interweaving Metric and Momentum to Navigate Flat Energy Landscapes

Variational quantum algorithms

Greedy Gradient-free Adaptive Variational Quantum Algorithms on a Noisy Intermediate Scale Quantum Computer