Anomaly Detection Using Deep Neural Network Quantum Encoder

Abstract

Quantum computers are replacing the existing classical computers in the very near future time. Due to the evolution of Quantum computers, machine learning techniques are used as a tool for recognizing patterns in the given data. Network security is a major issue nowadays. Many machine learning-based Auto encoders exist for anomaly detection in the network. The Deep learning neural networks based auto encoders increase the accuracy of the anomaly detection models. To implement various machine learning models on a quantum computer the classical data should be converted into qubits. Hence in this paper, we propose a deep neural network-based quantum auto encoder for detecting various network anomalies like Dos, Shellcode, Worm, and Backdoor. Dimensionality reduction is applied to reduce the size of the sample data features set. The proposed method yields high accuracy of 100% during the training phase and 99% during the testing phase.