A Hybrid Modeling of Mobile App Dynamics on Serial Causality for Malware Detection

Abstract

The popularity of smart phones has brought significant convenience to people’s lives, but also there are many security problems. In recent years, malicious applications are increasingly rampant, which threaten users and society as security challenges to network reliability and management. However, due to neglecting the sequential features between network flows, existing malicious application recognition methods based on network traffic analysis have low recognition accuracy. Based on the network traffic characteristics of Android applications, this paper firstly applies Long Short-Term Memory network-based variational Auto-Encoder to extract the sequential feature of the application running time. Then, we design the BP neural network for initial classification and connect the class vector output of the BP neural network with the original data. The output is fed into the cascade forest for further feature learning and classification. The integrated methods are easy to implement with data independency and efficiency. We conduct experiments to evaluate the proposed with Android malware dataset CICAndMal2017, with a 97.29% high accuracy, comparatively significant precision and recall rates when benchmarked against other methods.

Malware Detection in Android IoT Systems Using Deep Learning

“…The results also demonstrate that the proposed hybrid model outperforms other frameworks in terms of reliability. The framework [18] employed the Variational Auto Encoder (VAE) based LSTM model for selecting the feature, performing the training, and estimating the results. This scheme is effective only for sequential data and requires high computation power for processing.…”

“…This study achieved accuracy up to 99.2% but was not considered to detect the embedded or bound malware including file infector, riskware, rootkits and adware. The model proposed in [18] categorizes the detection module into two layers and achieves an accuracy of 98.29%. However, the main limitation of this work is the high computational power required to process the dual layers of data.…”

Detection of Android Malware Based on Deep Forest and Feature Enhancement

“…was used as classifier, the binary classification accuracy on CICAndMal2017 dataset was 98%. Liu [12] et al proposed an Android malware detection method based on the statistical characteristics of Android application communication traffic. This method applied LSTM-based variational Auto-Encoder (LSTM-VAE) to extract the time sequence characteristics.…”

“…Finally, considering the number of samples of each class is just about 100, the experiment also prove that this method is suitable for small-scale classification task. In order to further verify the effectiveness of FE-CaDF, the binary classification methods proposed in reference [11,12,13,14] are compared with it on CICAndMal2017. Table 14 shows the comparison results.…”

Improved AlexNet with Inception-V4 for Plant Disease Diagnosis

“…In [ 10 ], Triki et al proposed a leaf detection and segmentation model, deep leaf, which was based on Mask-RCNN and used morphological characteristics in plant specimens. Liu et al applied a long short-term memory network-based variational autoencoder to extract the sequential feature of the application running time [ 11 ]. Rao et al used bilinear convolutional neural networks (bi-CNNs) for identifying different types of leaves, where VGG and ResNet were used as feature extractors [ 12 ].…”