This paper presents three novel quantum auto-encoder-based anomaly detection frameworks for network security. The frameworks integrate quantum autoencoders with quantum one-class support vector machines, quantum random forests, and quantum k-nearest neighbor approaches. The goal is to develop hybrid models that combine the strengths of quantum and deep learning for efficient anomaly recognition. The anomaly detection capability of the frameworks is evaluated using benchmark datasets comprising computer and Internet of Things network flows. The results show that all three frameworks have high potential for accurately detecting network traffic anomalies, with the framework combining quantum autoencoder and quantum k-nearest neighbor achieving the highest accuracy. This demonstrates the promising potential of quantum frameworks for anomaly detection, highlighting their relevance for future advancements in network security. Quantum computing, with its unique attributes such as parallelism, offers high potential for many applications, including image recognition, protein folding, and fraud detection. Quantum machine learning (QML) has seen rapid development in pattern recognition, but its ability to enhance anomaly detection in network security is not well studied. The paper explores the current state of QML in the context of cyber security anomaly detection, specifically focusing on network traffic. It proposes quantum algorithms to improve the detection of anomalies from network traffic information obtained from computer or IoT networks. The paper also introduces three novel frameworks for anomaly detection using QML and QDL in conjunction with autoencoders. The frameworks are distinguished by innovative technical methods, with a strong emphasis on encoding strategies. The evaluation using NISQ quantum computers and IBM quantum simulators reveals that all three proposed frameworks improve the anomaly detection performance compared to classical counterparts on benchmark datasets. In particular, the framework combining quantum autoencoder and quantum kNN performs the best among the three.This paper presents three novel quantum auto-encoder-based anomaly detection frameworks for network security. The frameworks integrate quantum autoencoders with quantum one-class support vector machines, quantum random forests, and quantum k-nearest neighbor approaches. The goal is to develop hybrid models that combine the strengths of quantum and deep learning for efficient anomaly recognition. The anomaly detection capability of the frameworks is evaluated using benchmark datasets comprising computer and Internet of Things network flows. The results show that all three frameworks have high potential for accurately detecting network traffic anomalies, with the framework combining quantum autoencoder and quantum k-nearest neighbor achieving the highest accuracy. This demonstrates the promising potential of quantum frameworks for anomaly detection, highlighting their relevance for future advancements in network security. Quantum computing, with its unique attributes such as parallelism, offers high potential for many applications, including image recognition, protein folding, and fraud detection. Quantum machine learning (QML) has seen rapid development in pattern recognition, but its ability to enhance anomaly detection in network security is not well studied. The paper explores the current state of QML in the context of cyber security anomaly detection, specifically focusing on network traffic. It proposes quantum algorithms to improve the detection of anomalies from network traffic information obtained from computer or IoT networks. The paper also introduces three novel frameworks for anomaly detection using QML and QDL in conjunction with autoencoders. The frameworks are distinguished by innovative technical methods, with a strong emphasis on encoding strategies. The evaluation using NISQ quantum computers and IBM quantum simulators reveals that all three proposed frameworks improve the anomaly detection performance compared to classical counterparts on benchmark datasets. In particular, the framework combining quantum autoencoder and quantum kNN performs the best among the three.