This study focuses on the anomaly detection problem in Network Security Situational Awareness (NSSA). We systematically review traditional approaches and recent advancements based on Machine Learning (ML) and Deep Learning (DL), and assess the application status of various anomaly detection models within the context of NSSA. Building upon this comprehensive review, we identify and analyze the primary challenges currently faced by the field, including the technical complexity of multimodal ML data fusion, insufficient real-time adaptive detection capabilities, limited model interpretability, and bottlenecks in privacy protection and security collaboration. To address these issues, we propose an integrated conceptual model, AFEMAD, which incorporates multimodal ML data fusion, online adaptive detection, enhanced interpretability, Federated Learning (FL), and collaborative privacy-preserving mechanisms, thereby providing an innovative technical framework for anomaly detection in NSSA. Finally, this study deepens the understanding of the anomaly detection problem in NSSA and provides both theoretical and practical guidance for the development of efficient and scalable anomaly detection systems.
As the Internet of Things (IoT) becomes increasingly indispensable across various domains, the connectivity between humans, machines, and devices intensifies. With the surge in IoT devices deployed in numerous fields, security risks have escalated, particularly the proliferation of botnets. Due to IoT devices’ limited resources and performance, they are easily exploited by malware to form part of IoT botnets. Numerous studies have attempted to detect IoT botnets using artificial intelligence, yielding promising results. However, challenges such as poor interpretability, high false positive rates, and difficulty in adapting to new and evolving threats persist. This paper presents a systematic literature review of IoT botnet detection research conducted over the past five years, initially filtering 143 studies and ultimately reviewing 134 based on a quality assessment. It examines the IoT botnet lifecycle, which describes the formation of a botnet, and evaluates various detection methods, with a focus on the types of features extracted for detection and how they relate to different phases of the botnet lifecycle. Additionally, an IoT botnet malicious activity map is presented to address the challenges faced by current detection models. This study also highlights key limitations and future research directions, offering valuable insights to improve the performance of IoT botnet detection systems.
5 December 2017 in Multimedia Tools and Applications
Facial Expression Recognition (FER) on unconstrained datasets poses a significant challenge, primarily due to data uncertainty stemming from human subjectivity and ambiguous facial expressions. Previous methods attempt to address this issue through relabeling strategies. However, this work reveals a relabel inconsistency problem. Specifically, the model weights are not updated simultaneously with the relabeling process. Consequently, the feature representations of the noisy samples remain associated with the previous label despite being relabeled. As a result, the relabeling mechanism reverts the new label back to the previous one, initiating a cycle between the two classes during the subsequent training. The failure to “shift” the feature representations closer to the new label centers hinders the model from learning discriminative features capable of handling data uncertainty, leading to degraded performance. In this work, a new framework based on embedding proximity is proposed to ensure consistent updating of feature representations with rectifications made during relabeling to overcome this limitation. This is achieved by pushing relabeled images closer to their newly assigned class centers and farther away from their previous class (wrong) centers in the feature embedding space. Through comprehensive experiments, this work utilizes existing models-SCN, RUL, and DMUE-to map the original feature space and then applies the proposed embedding proximity technique to update the feature representations. The updated models, denoted as SCN-C, RUL-C, and DMUE-C, demonstrate significant improvements in addressing inconsistency issues and enhancing overall performance. The proposed models outperform state-of-the-art methods, achieving accuracies of 65.73% on AffectNet, 89.51% on RAF-DB, and 71.83% on FER2013.