Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Quantum Enhanced Anomaly Detection for ADS-B Data using Hybrid Deep Learning

Rani Naaman, Felipe Gohring de Magalhaes, Jean-Yves Ouattara, Gabriela Nicolescu

TL;DR

The paper investigates quantum-enhanced anomaly detection for ADS-B data by proposing a hybrid quantum-classical neural network (H-FQNN) and comparing it to a classical FNN on a public ADS-B dataset. It implements a variational quantum classifier as the quantum layer and evaluates two loss functions, BCEWithLogitsLoss and CrossEntropyLoss, using six features derived from trajectory data. Results show that the H-FQNN achieves competitive accuracy (90.17%–94.05%) and F1 scores up to 93.99%, with six qubits often providing optimal performance, closely matching or slightly surpassing the classical baseline in several configurations. The study demonstrates the feasibility of quantum-assisted anomaly detection in aviation data and outlines practical considerations for architecture tuning and hardware deployment, highlighting the need for real-qubit experiments and broader benchmarking in future work.

Abstract

The emerging field of Quantum Machine Learning (QML) has shown promising advantages in accelerating processing speed and effectively handling the high dimensionality associated with complex datasets. Quantum Computing (QC) enables more efficient data manipulation through the quantum properties of superposition and entanglement. In this paper, we present a novel approach combining quantum and classical machine learning techniques to explore the impact of quantum properties for anomaly detection in Automatic Dependent Surveillance-Broadcast (ADS-B) data. We compare the performance of a Hybrid-Fully Connected Quantum Neural Network (H-FQNN) with different loss functions and use a publicly available ADS-B dataset to evaluate the performance. The results demonstrate competitive performance in detecting anomalies, with accuracies ranging from 90.17% to 94.05%, comparable to the performance of a traditional Fully Connected Neural Network (FNN) model, which achieved accuracies between 91.50% and 93.37%.

Quantum Enhanced Anomaly Detection for ADS-B Data using Hybrid Deep Learning

TL;DR

The paper investigates quantum-enhanced anomaly detection for ADS-B data by proposing a hybrid quantum-classical neural network (H-FQNN) and comparing it to a classical FNN on a public ADS-B dataset. It implements a variational quantum classifier as the quantum layer and evaluates two loss functions, BCEWithLogitsLoss and CrossEntropyLoss, using six features derived from trajectory data. Results show that the H-FQNN achieves competitive accuracy (90.17%–94.05%) and F1 scores up to 93.99%, with six qubits often providing optimal performance, closely matching or slightly surpassing the classical baseline in several configurations. The study demonstrates the feasibility of quantum-assisted anomaly detection in aviation data and outlines practical considerations for architecture tuning and hardware deployment, highlighting the need for real-qubit experiments and broader benchmarking in future work.

Abstract

The emerging field of Quantum Machine Learning (QML) has shown promising advantages in accelerating processing speed and effectively handling the high dimensionality associated with complex datasets. Quantum Computing (QC) enables more efficient data manipulation through the quantum properties of superposition and entanglement. In this paper, we present a novel approach combining quantum and classical machine learning techniques to explore the impact of quantum properties for anomaly detection in Automatic Dependent Surveillance-Broadcast (ADS-B) data. We compare the performance of a Hybrid-Fully Connected Quantum Neural Network (H-FQNN) with different loss functions and use a publicly available ADS-B dataset to evaluate the performance. The results demonstrate competitive performance in detecting anomalies, with accuracies ranging from 90.17% to 94.05%, comparable to the performance of a traditional Fully Connected Neural Network (FNN) model, which achieved accuracies between 91.50% and 93.37%.

Paper Structure

This paper contains 16 sections, 12 equations, 9 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Basic Concepts
  3. Quantum Computing (QC)
  4. Quantum Machine Learning (QML)
  5. Variational Quantum Classifier (VQC)
  6. State-of-the-art
  7. Methodology
  8. Model #1: Hybrid-Fully Connected Quantum Neural Network (H-FQNN)
  9. Model #2: Fully Connected Neural Network (FNN)
  10. DATASET
  11. Feature Selection
  12. Data Preprocessing
  13. Evaluation Matrix
  14. Results
  15. Discussion
  16. ...and 1 more sections

Figures (9)

  • Figure 1: Four different approaches to combine the disciplines of quantum computing and machine learning. (The figure is adapted from Schuld and Petruccione Schuld2021)
  • Figure 2: An example of a VQC architecture with 6 qubits, strongly entangling layers and Pauli-Z basis measurement
  • Figure 3: Quantum Neural Network Architecture
  • Figure 4: Neural Network Architecture
  • Figure 5: Correlation matrix — Feature to label.
  • ...and 4 more figures