Abstract-Network Intrusion Detection Systems (NIDSs) play a crucial role in defending computer networks. However, there are concerns regarding the feasibility and sustainability of current approaches when faced with the demands of modern networks. More specifically, these concerns relate to the increasing levels of required human interaction and the decreasing levels of detection accuracy. This paper presents a novel deep learning technique for intrusion detection, which addresses these concerns. We detail our proposed non-symmetric deep auto-encoder (NDAE) for unsupervised feature learning. Furthermore, we also propose our novel deep learning classification model constructed using stacked NDAEs. Our proposed classifier has been implemented in GPU-enabled TensorFlow and evaluated using the benchmark KDD Cup '99 and NSL-KDD datasets. Promising results have been obtained from our model thus far, demonstrating improvements over existing approaches and the strong potential for use in modern NIDSs.

No abstract

No abstract

Anomaly-based Network Intrusion Detection Systems (NIDSs) are a common security defense for modern networks. The success of their operation depends upon vast quantities of training data. However, one major limitation is the inability of NIDS to be reliably trained using imbalanced datasets. Network observations are naturally imbalanced, yet without substantial data pre-processing, NIDS accuracy can be significantly reduced. With the diversity and dynamicity of modern network traffic, there are concerns that the current reliance upon unnatural balanced datasets cannot remain feasible in modern networks. This paper details our de-noising method, which when combined with deep learning techniques can address these concerns and offer accuracy improvements of between 1.5% and 4.5%. Promising results have been obtained from our model thus far, demonstrating improvements over existing approaches and the strong potential for use in modern NIDSs.

Despite the continued advancements in security research, malware persists as being a major threat in this digital age. Malware detection is a primary defence strategy for most networks but the identification of malware strains is becoming increasingly difficult. Reliable identification is based upon characteristic features being detectable within an object. However, the limitations and expense of current malware feature extraction methods is significantly hindering this process. In this paper, we present a new method for identifying malware based on behavioural feature extraction. Our proposed method has been evaluated using seven classification methods whilst analysing 2,068 malware samples from eight different families. The results achieved thus far have demonstrated promising improvements over existing approaches.

No abstract