Abstract:In this article we present the application of classical and quantum-classical hybrid anomaly detection schemes to explore exotic configuration with anomalous features. We consider the Anderson model as a prototype where we define two types of anomalies - a high conductance in presence of strong impurity and low conductance in presence of weak impurity - as a function of random impurity distribution. Such anomalous outcome constitutes an imperceptible fraction of the data set and is not a part of the training process. These exotic configurations, which can be a source of rich new physics, usually remain elusive to conventional classification or regression methods and can be tracked only with a suitable anomaly detection scheme. We also present a systematic study of the performance of the classical and the quantum-classical hybrid anomaly detection method and show that the inclusion of a quantum circuit significantly enhances the performance of anomaly detection which we quantify with suitable performance metrics. Our approach is quite generic in nature and can be used for any system that relies on a large number of parameters to find their new configurations which can hold exotic new features.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to explore how to use classical and quantum - classical hybrid anomaly detection methods in deep learning to identify strange configurations with abnormal features. Specifically, by studying the Anderson model, the author defines two types of anomalies: 1. **High conductivity in the presence of strong impurities**: In the case of strong impurity scattering, the system usually exhibits low conductivity, but some special configurations may lead to significantly higher conductivity. 2. **Low conductivity in the presence of weak impurities**: In the case of weak impurity scattering, the system usually exhibits high conductivity, but some special configurations may lead to significantly lower conductivity. These abnormal configurations account for a very small proportion in the data set, and these abnormal data are not included in the training process. Therefore, traditional classification or regression methods are difficult to discover these strange configurations. However, these configurations may contain new physical phenomena and are of great significance for device design and materials science. ### Main research content 1. **Model selection**: - Use three methods, namely the classical Isolation Forest (IF), Autoencoder (AE), and Hybrid Quantum - Classical Autoencoder (HAE), for anomaly detection. 2. **Performance evaluation**: - Quantify the performance of different methods through indicators such as Precision, Recall, and F1 - Score. - The results show that HAE performs better than its classical counterpart (AE) in dealing with highly nonlinear feature mappings, which is attributed to the ability of the quantum circuit (PQC) to process data in high - dimensional Hilbert space. 3. **Application prospects**: - These anomaly detection schemes can be used to understand Anderson localization behavior and the formation of solitons in disordered systems. - The methods are general and can be extended to other systems, such as the study of light localization in optical lattices or the detection of novel configurations in higher - dimensional phase spaces. ### Formula display The formulas involved in the paper include: - **Zero - bias conductivity**: \[ T=\text{Tr}\left[\Gamma_1 G^R \Gamma_2 G^A\right] \] where \(G^{R,A}=\left[E - H-\Sigma^{R,A}_1-\Sigma^{R,A}_2\right]^{-1}\) are the complex/advanced Green functions of the scattering region, and \(\Gamma_{1,2}=i\left[\Sigma^R_{1,2}-\Sigma^A_{1,2}\right]\) are the complex/advanced self - energies of the left and right electrodes. - **Mean - squared - error loss**: \[ \text{MSE}=\frac{1}{n}\sum_{i = 1}^n(x_i-\hat{x}_i)^2 \] where \(x\) and \(\hat{x}\) are the original data points and the predicted data points respectively. - **Performance evaluation indicators**: \[ \text{Precision}=\frac{TP}{TP + FP},\quad \text{Recall}=\frac{TP}{TP + FN} \] \[ F1\text{-score}=\frac{2\times\text{Precision}\times\text{Recall}}{\text{Precision}+\text{Recall}} \] where TP, FP, and FN represent the numbers of true positives, false positives, and false negatives respectively. In conclusion, this paper successfully reveals some in the Anderson model that are difficult to be discovered by traditional methods by introducing classical and quantum - classical hybrid anomaly detection methods.

Exploring exotic configurations with anomalous features using deep learning: Application of classical and quantum-classical hybrid anomaly detection

Long-lived Particles Anomaly Detection with Parametrized Quantum Circuits

Quantum Machine Learning Algorithms for Anomaly Detection: a Review

A Quantum-Classical Hybrid Solution for Deep Anomaly Detection

Quantum anomaly detection in the latent space of proton collision events at the LHC

Unravelling physics beyond the standard model with classical and quantum anomaly detection

Exploring Unsupervised Anomaly Detection with Quantum Boltzmann Machines in Fraud Detection

Quantum similarity learning for anomaly detection

Detecting Quantum Anomalies in Open Systems

Unsupervised Quantum Anomaly Detection on Noisy Quantum Processors

Quantum-inspired anomaly detection, a QUBO formulation

Quantum multi-anomaly detection

Adversarial attacks on hybrid classical-quantum Deep Learning models for Histopathological Cancer Detection

Quantum Generative Adversarial Networks For Anomaly Detection In High Energy Physics

Variational Quantum Anomaly Detection: Unsupervised mapping of phase diagrams on a physical quantum computer

Quantum Machine Learning for Anomaly Detection in Consumer Electronics

Transfer Learning Based Hybrid Quantum Neural Network Model for Surface Anomaly Detection

Machine learning on quantum experimental data toward solving quantum many-body problems

Unsupervised event classification with graphs on classical and photonic quantum computers