Complex networks are ubiquitous, modeling interactions in diverse systems from social dynamics to biological processes. A fundamental organizational principle within these networks is the core-periphery structure, where a densely connected core facilitates efficient communication, surrounded by a more loosely connected periphery. Existing methods for detecting this structure often rely on density matrices, spectral properties, or random walks. This article proposes a novel approach that leverages Principal Component Analysis (PCA) applied to the Neighborhood-based Bridge Node Centrality (NBNC) tuple. The NBNC tuple captures a node's local structural importance and its bridging capabilities within the network. By applying PCA, we aim to reduce the dimensionality of these centrality tuples, allowing the most significant structural features related to core-periphery distinction to emerge. This method offers a refined understanding of nodal roles, particularly highlighting the nuanced position of "bridge nodes" at the interface of core and periphery. Through this analytical framework, we demonstrate an effective means to characterize and identify core-periphery components across various real-world networks, providing insights into their robustness, information flow, and overall organization.

Delay-Tolerant Networks (DTNs) are designed to operate effectively in environments characterized by frequent disconnections, long delays, and intermittent connectivity. While their store-carry-forward paradigm enables communication in challenged environments, it also introduces unique security vulnerabilities, particularly concerning attacks that exploit spatial and temporal patterns of node mobility and contact. This article proposes and explores the feasibility of a geospatial anomaly detection framework to identify and monitor potential attack locations within a specific area of a DTN. By leveraging location information alongside network performance metrics, this approach aims to proactively detect malicious activities, such as black hole attacks or resource exhaustion, confined to geographical regions. The methodology encompasses data collection, feature engineering combining network and spatial data, and the application of anomaly detection algorithms. The hypothetical results suggest that such a system could significantly enhance DTN security by enabling targeted intervention and improving overall network resilience in challenged communication scenarios.

Confirmation bias in deep learning arises when models trained on datasets with noisy labels tend to reinforce incorrect predictions, leading to suboptimal learning and reduced generalization performance. This paper proposes a collaborative network training framework to mitigate confirmation bias in the presence of label noise. In the proposed method, two networks are trained simultaneously, each selecting clean samples for the other to learn from. This cross-training strategy prevents individual networks from overfitting to noisy labels and helps preserve model diversity. The framework also incorporates a sample agreement mechanism and consistency regularization to further stabilize training and improve robustness. Experimental evaluations on benchmark datasets including CIFAR-10, CIFAR-100, and Clothing1M show that the proposed approach outperforms existing noise-robust training methods, achieving higher accuracy and better noise tolerance. The results validate the effectiveness of collaborative learning in reducing confirmation bias and improving model reliability under label noise.

Malware continues to pose a significant threat to cybersecurity, evolving rapidly in complexity and evasion techniques. Traditional detection methods often struggle against sophisticated attacks due to their reliance on static signatures or limited understanding of program behavior. This article introduces a novel dynamic malware detection approach that leverages both function parameter encoding and function dependency modeling derived from Application Programming Interface (API) call sequences. By capturing the rich contextual information conveyed through API call parameters and understanding the intricate relationships between function invocations, our method aims to provide a more robust and accurate classification of malicious software. We detail the methodology, from dynamic analysis and data collection to the feature engineering and model training, and present results demonstrating superior performance compared to existing techniques that primarily rely on API call sequences alone. The findings underscore the importance of deeper behavioral analysis for effective malware detection in the contemporary threat landscape.

This article presents the design, implementation, and validation of a real-time digital twin for a Stewart platform, integrated with advanced trajectory computation capabilities. Stewart platforms, known for their high precision and multi-degree-of-freedom motion, are widely used in applications such as flight simulators, medical devices, and manufacturing systems. The integration of a digital twin allows for real-time monitoring, predictive analysis, and enhanced control, thereby improving the system's operational efficiency and resilience. This work details the architectural framework of the digital twin, the methods for real-time data synchronization, the algorithms for dynamic trajectory generation, and the validation of the virtual model against the physical Stewart platform. The proposed system demonstrates significant potential for improving the control, diagnostics, and overall performance of complex robotic manipulators through a comprehensive virtual representation.