Abstract:The increasing capabilities of Machine Learning (ML) models go hand in hand with an immense amount of data and computational power required for training. Therefore, training is usually outsourced into HPC facilities, where we have started to experience limits in scaling conventional HPC hardware, as theorized by Moore's law. Despite heavy parallelization and optimization efforts, current state-of-the-art ML models require weeks for training, which is associated with an enormous $CO_2$ footprint. Quantum Computing, and specifically Quantum Machine Learning (QML), can offer significant theoretical speed-ups and enhanced expressive power. However, training QML models requires tuning various hyperparameters, which is a nontrivial task and suboptimal choices can highly affect the trainability and performance of the models. In this study, we identify the most impactful hyperparameters and collect data about the performance of QML models. We compare different configurations and provide researchers with performance data and concrete suggestions for hyperparameter selection.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to optimize the hyper - parameters of quantum neural networks (QNNs) to improve their training efficiency and performance. Specifically, the research mainly addresses the following key issues: 1. **Limitations of classical machine - learning models**: - Classical machine - learning models require a large amount of data and computing resources during training, resulting in long training times and high carbon emissions. - Training is usually outsourced to high - performance computing (HPC) facilities, but these facilities also face expansion bottlenecks, as predicted by Moore's Law. 2. **Advantages and challenges of quantum machine learning (QML)**: - Quantum computing, especially quantum machine learning, can theoretically provide significant speed - ups and stronger expressive power. - However, training QML models requires adjusting multiple hyper - parameters, and inappropriate choices can seriously affect the trainability and performance of the model. 3. **Specific problems of hyper - parameter optimization**: - Identify the hyper - parameters that have the greatest impact on QNN performance and collect data on the performance of QML models. - Compare performance under different configurations and provide specific hyper - parameter selection suggestions for researchers. 4. **The impact of hardware noise**: - Study the impact of hardware noise on different hyper - parameter settings, especially under the currently available Noisy Intermediate - Scale Quantum (NISQ) technology. 5. **Optimization strategies**: - Provide specific selection suggestions for hyper - parameters such as optimizers and initialization methods to overcome problems such as the Barren Plateau phenomenon, thereby improving the training effect of QNNs. ### Specific contributions - **Data collection and analysis**: Performance data of QML models under different hyper - parameter configurations were collected through experiments. - **Hyper - parameter impact assessment**: The impact of optimizers, initialization methods, and entanglement strategies on QNN performance was evaluated. - **Hardware noise research**: The impact of hardware noise on different hyper - parameter settings was studied. - **Open - source code and results**: All code and results have been made public for easy reproducibility and further research by other researchers. ### Summary Through a systematic study of the hyper - parameters of quantum neural networks, this paper aims to provide researchers with effective methods and guidance for optimizing QNN training, thereby promoting the development of quantum machine learning.

On Optimizing Hyperparameters for Quantum Neural Networks

Learning To Optimize Quantum Neural Network Without Gradients

Quantum Algorithms for Feedforward Neural Networks

Hyperparameter optimization, quantum-assisted model performance prediction, and benchmarking of AI-based High Energy Physics workloads using HPC

QuaLITi: Quantum Machine Learning Hardware Selection for Inferencing with Top-Tier Performance

Sparks of Quantum Advantage and Rapid Retraining in Machine Learning

QNet: A Scalable and Noise-Resilient Quantum Neural Network Architecture for Noisy Intermediate-Scale Quantum Computers

Training Classical Neural Networks by Quantum Machine Learning

Machine learning the computational cost of quantum chemistry

Integrated Encoding and Quantization to Enhance Quanvolutional Neural Networks

Studying the Impact of Quantum-Specific Hyperparameters on Hybrid Quantum-Classical Neural Networks

An Empirical Study of Optimizers for Quantum Machine Learning

A Quantum Circuit-Based Compression Perspective for Parameter-Efficient Learning

Optimizing Temperature Distributions for Training Neural Quantum States using Parallel Tempering

A Study on Optimization Techniques for Variational Quantum Circuits in Reinforcement Learning

Quantum Machine Learning hyperparameter search

Quantum inspired high dimensional hyperparameter optimization of machine learning model

Quantum-Train: Rethinking Hybrid Quantum-Classical Machine Learning in the Model Compression Perspective

A Hyperparameter Study for Quantum Kernel Methods

Distributed hybrid quantum-classical performance prediction for hyperparameter optimization

A Post-Training Approach for Mitigating Overfitting in Quantum Convolutional Neural Networks