Search

In this paper, a new deep learning and machine learning combined model is proposed for malware behavior analysis. One part of it analyzes the dependency relation in API (Application Programming Interface) call sequence at the functional level, and extracts features for random forest to learn and classify. The other part employs a bidirectional residual neural network to study the API sequence and discover malware with redundant information preprocessing. In the API call sequence, future information is much more important for conjecturing the semantic of the current API call. We conducted experiments on a malware dataset. The experiment results show that both methods can effectively detect malwares. However, the combined framework has better classification performance. The classification accuracy of the combined malware detection architecture is 0.967.

Personal keystroke mode is difficult to imitate and can therefore be used for identity authentication. According to the keystroke data when a person inputs free text, the keystroke habit of the person can be learned. Detecting a user’s keystroke habits as the user enters text can continuously verify the user’s identity without affecting user input. This paper proposes to divide the user keystroke data into a fixed-length keystroke sequence, and convert the keystroke sequence into a keystroke vector sequence according to the time feature of the keystroke. A model of a recursive neural network plus a convolutional neural network is used to learn a sequence of individual keystroke vectors to obtain individual keystroke features for identity authentication. The model was tested using an open data set and the best False Rejection Rate(FRR) was 1.95%, False Acceptance Rate (FAR) was 4.12% and Equal Error Rate(EER) was 3.04%.

Dynamic analysis of malware sample is an important method in the malware detection. In this paper, a malware detection architecture is proposed that combines machine learning and deep learning. The combination classification architecture focuses on the dynamic behavior of a malware sample. A new feature extraction method is proposed for dynamic behavior analysis. In this paper, recurrent neural network model is used to extracts the abstract features. Several sequence data preprocessing methods are studied to remove the redundant data. Experiments show that the AUC of the combination architecture is 99.3%. The classification performance of the combination architecture is better than the separate machine learning or deep learning.

Constant advances in Micro Electro-Mechanical System (MEMS) technology promote miniaturization, low-cost and energy-efficiency of smart sensors. Meanwhile, Wireless Sensor Networks (WSNs) emerge as a collaborative network composed of such numerous sensors via wireless communication integrated. WSNs have a wide range of applications in the physical world, such as military surveillance, target tracking, healthcare etc. This paper illustrates a comprehensive study we contribute to in overviews, design ingredients, protocol architecture, the current research focus and the development status for WSNs with a contemporary view.