AI-centric Proxy Design Synthesis for Non-Obvious Link/Entity and Higher-Order Network Discernment

Chan, Steve

Home // International Journal On Advances in Intelligent Systems, volume 17, numbers 3 and 4, 2024 // View article

AI-centric Proxy Design Synthesis for Non-Obvious Link/Entity and Higher-Order Network Discernment

Authors:
Steve Chan

Keywords: Intelligent Decision-Making Systems; Artificial Intelligence; Machine Learning; Big Data; Advanced Analytics; Command and Control; Large Scale Complex Networks.

Abstract:
The discernment of relevant sparse and “Very Small/Non-Obvious” (VSNO) clusters within High Dimensional Data (HDD) and the operationalization of Spatio- Temporal Knowledge Graph Completion (STKGC) for High- Order Network (HON) discernment are NP-Hard. The amalgam of a Lower Ambiguity, Higher Uncertainty (LAHU)/Higher Ambiguity, Lower Uncertainty (HALU) Module (LHM), Isomorphic Paradigm (IsoP) Comparator Similarity Measure Module (ICSM2), Multi-Criteria Decision- Making Module (MCDM2), Information Fusion Module (IFM), AI Energy Consumption Module (AECM), and a bespoke Metaheuristic Algorithm Module (MAM) are delineated in this paper to show the potentiality for the concurrent treatment of VSNO, STKGC, and HON, which are essential for Advanced Analytic Technologies (AAT)/Advanced Anomaly Detection (AAD), At-the-Edge Observational Space Analysis (AOSA), and Continuous Situational Awareness (CSA). These are vital aspects for critical applications, such as, among others, network analysis (e.g., C2 systems) and maritime domain awareness. The described LHM-ICSM2- MCDM2-IFM-AECM-MAM amalgam can be operationalized by a bespoke Graph Convolutional Network (GCN)- Bidirectional Long Short-Term Memory (BiLSTM)-Graph- Attention-Network (GAT) mechanism along with a Robust Convex Relaxation (RCR)-based Deep Convolutional [Neural Network] Generative Adversarial Network (DCGAN)- Hypergraph-Induced Infimal Convolutional Manifold Neural Network (H-IICMNN)-1,2,3,4 architectural construct (GCN- BiLSTM-GAT & RCR-DCGAN-[H-IICMNN]-1,2,3,4 or GBGRDH-1,2,3,4) to address the involved NP-Hard problems.

Pages: 158 to 177

Publication date: December 30, 2024

Published in: journal

ISSN: 1942-2679