Preventing Data Hallucinations in Agentic AI for Optical Network Troubleshooting

arXiv:2610.23252 · cs.NI, cs.AI · Submitted 2026-10-16 · Read on arXiv

cs.NI, cs.AI

Submitted: 2026-10-16

Updated: 2026-10-16

Comments: Presented as a demo at the 52nd European Conference on Optical Communication (ECOC 2026)

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Agentic AI — 1× in this paper · explained in 2 episodes
• Hallucination — 1× in this paper · explained in 4 episodes

Related papers

• HiFiNet: Hierarchical Fault Identification in Wireless Sensor Networks via Edge-Based Classification and Graph Aggregation
• PACC: Protocol-Aware Cross-Layer Compression for Compact Network Traffic Representation
• TraceCodec: A Compiler-Backed Neural Codec for Network Traffic Traces
• Embodied AI in 6G Networks: From Intelligent Connectivity to Physical Intelligence
• Lightweight GenAI for Network Traffic Generation: Fidelity, Augmentation, and Classification
• NetRL: Powering Network Decisions via NetGRPO-Driven Reinforcement Learning