Image Based Malware Detection Using Transfer Learning and Hybrid Quantum Machine Learning Models

Image based malware classification is getting more popular in the field of malware detection as it frees the researcher from the tedious process disassembling and analyzing the code of the malicious software. Deep learning and transfer learning methods are efficient for extracting features from images. The accuracy of prediction will be better with the quality of the features extracted. This paper carries out experiments on a benchmark dataset, Malimg using the deep learning model CNN and transfer learning models VGG16 and ResNet. Within the broader paradigm of Quantum Computing, one of the primary research areas is Quantum Machine Learning (QML). Since applying QML algorithms to solve real-world problems might potentially save time and money when compared to more traditional (or digital) machine learning methods, researchers have become quite interested in QML in recent years. We also developed a HQCCNN which is a classical and quantum hybrid version of convolutional neural network for classifying the images that integrates classical and quantum counterparts. A parameterized quantum circuit is utilized in the design of the quantum convolutional layer. To retrieve hidden informatio