This research paper explores an adversarial bandit problem for quantum network routing using the Ekert-91 protocol. The study models a scenario where Alice selects a route and Eve attempts to degrade the network through various attack surfaces. By employing adversarial co-learning across 50 topologies, the paper demonstrates that learned retention closely mirrors a minimax reference, with bottleneck routes showing zero retention and non-bottleneck routes following a coverage principle. The findings are further analyzed using decision-tree explanation models, leading to an open-source workflow for summarizing evidence in quantum-repeater network games. AI
IMPACT This research explores adversarial learning techniques applicable to quantum networks, potentially influencing future AI-driven network optimization strategies.
RANK_REASON The cluster contains an academic research paper published on arXiv.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
