Graph-optimized Neural Networks with Topological Embeddings for Scalable Cyber Threat Detection in Internet of Things Network

The Internet of Things (IoT) connects programs, devices, data storage, and services that, since they always offer services to the business, may create new opportunities for cyber-attacks. At the moment, there is a considerable risk of malware attacks and software piracy endangering IoT security. These risks can potentially steal crucial information that harms the company’s financial reputation. Hence, this study proposes a novel graph-optimized neural network (GONN) with topological embedding for anomaly and cyber threat identification. The data is transformed into a common scale using the min-max normalization approach. The suggested approach enhances feature representation by integrating graph topology-aware learning, allowing for improved classification of normal and abnormal activities in large-scale networks. The technique guarantees Scalability and computing efficiency, making it a good fit for real-time threat monitoring. Experimental tests on standard cybersecurity datasets show that GONN reduces false positives, is more computationally efficient, and has better detection accuracy than conventional deep learning algorithms. The findings demonstrate that graph-based optimization helps enhance cyber threat detection by providing a more resilient and adaptable security framework.