Due to the rapid rise of automated tools, the number of malware variants has increased dramatically, which poses a tremendous threat to the security of the Internet. Recently, some methods for quick analysis of malware have been proposed, but these methods usually require a large computational overhead and cannot classify samples accurately for large-scale and complex malware data set. Therefore, in this paper, we propose a new visualization method for characterizing malware globally and locally to achieve fast and effective fine-grained classification. We take a new approach to visualize malware as RGB-colored images and extract global features from the images. Gray-level co-occurrence matrix and color moments are selected to describe the global texture features and color features, respectively, which produces low-dimensional feature data to reduce the complexity of training model. Moreover, a series of special byte sequences are extracted from code sections and data sections of malware and are processed into feature vectors by Simhash as the local features. Finally, we merge the global features and local features to perform malware classification using random forest, K-nearest neighbor, and support vector machine. Experimental results show that our approach obtains the highest accuracy of 97.47% and the highest F-measure of 96.85% of 7087 samples from 15 families. Color features and the local features effectively assist in the classification based on texture features and enhance the F-measure by 3.4% and 1%, respectively. Overall, the combination of global features and local features can realize fine-grained malware classification with low computational cost.
18 October 2022 in Expert Systems with Applications
With the rapid evolution of the Internet, the application of artificial intelligence fields is more and more extensive, and the era of AI has come. At the same time, adversarial attacks in the AI field are also frequent. Therefore, the research into adversarial attack security is extremely urgent. An increasing number of researchers are working in this field. We provide a comprehensive review of the theories and methods that enable researchers to enter the field of adversarial attack. This article is according to the “Why? → What? → How?” research line for elaboration. Firstly, we explain the significance of adversarial attack. Then, we introduce the concepts, types, and hazards of adversarial attack. Finally, we review the typical attack algorithms and defense techniques in each application area. Facing the increasingly complex neural network model, this paper focuses on the fields of image, text, and malicious code and focuses on the adversarial attack classifications and methods of these three data types, so that researchers can quickly find their own type of study. At the end of this review, we also raised some discussions and open issues and compared them with other similar reviews.
Cloud-assisted smart grid has been broadly deployed to improve the economics, efficiency, sustainability, and reliability of electricity services. The new revolution of technology will also bring new challenges to data security, particularly data confidentiality. As a promising encryption method for supporting fine-grained access control, attribute-based encryption (ABE) is widely given attention by the academia and industry. However, most existing schemes suffer from the efficiency problem limiting its deployment for the cloud-assisted smart grid, especially for resource-limited users. To address this problem, this paper proposes two efficient pairing-free ciphertext-policy attribute-based schemes that eliminate the computation-intensive bilinear pairing operation. The basic scheme only uses some simple scalar multiplications based on elliptic curves, and the enhanced scheme provides an important usability feature, namely the users and attributes revocation. The security of the proposed schemes is reduced to the elliptic curve decisional Diffie-Hellman problem. Finally, the performance analysis and extensive experiments demonstrate that our proposed schemes are suitable for the cloud-assisted smart grid.
While deep learning techniques have been extensively employed in malware detection, there is a notable challenge in effectively embedding malware features. Current neural network methods primarily capture superficial characteristics, lacking in-depth semantic exploration of functions and failing to preserve structural information at the file level. Motivated by the aforementioned challenges, this paper introduces MalHAPGNN, a novel framework for malware detection that leverages a hierarchical attention pooling graph neural network based on enhanced call graphs. Firstly, to ensure semantic richness, a Bidirectional Encoder Representations from Transformers-based (BERT) attribute-enhanced function embedding method is proposed for the extraction of node attributes in the function call graph. Subsequently, this work designs a hierarchical graph neural network that integrates attention mechanisms and pooling operations, complemented by function node sampling and structural learning strategies. This framework delivers a comprehensive profile of malicious code across semantic, syntactic, and structural dimensions. Extensive experiments conducted on the Kaggle and VirusShare datasets have demonstrated that the proposed framework outperforms other graph neural network (GNN)-based malware detection methods.