As artificial intelligence techniques evolve, we are approaching a critical moment for the widespread deployment of autonomous vehicles. Subsequently, the emergence of mixed-autonomy traffic environments presents formidable challenges to autonomous vehicles, especially for the accurate prediction of lane change intentions of their surrounding human-driven vehicles, which is crucial for ensuring the safety of autonomous vehicles. Existing lane change prediction models mainly focus on capturing the temporal variations in the movement dynamics of individual vehicles. However, the neglect to consider inter-vehicle interactions hinders their capability in complex lane change scenarios, resulting in suboptimal prediction performance. Moreover, current interaction-aware approaches for autonomous driving fail to explicitly model future interactions between vehicles, leading to unreasonable prediction results that can cause collisions between vehicles. To address the above issues, we propose to incorporate the concept of perceived safety into future interaction modeling and design a dual-view interaction-aware lane change prediction model. We evaluate the proposed model on two real-world datasets and experimental results show that the proposed model achieves average improvements of 11.7-12.4% in classification ability and 75.6-95.7% in forecast ability over the best-performing baselines across the two datasets. The ablation study and investigation into future interaction modeling demonstrate that our model has advantages in interpreting lane change scenarios from a driving safety perspective.

Distributed network security requires multiple defenders to make coupled resource allocation and coordination decisions under limited budgets, connectivity-dependent risks, and non-negligible coordination costs. We develop a mathematical framework for jointly optimizing defensive resource allocation and coalition formation in a multi-defender network security game. Nodes have heterogeneous values and protection thresholds, attacks propagate through vulnerable connected components, and defenders retain individual budgets while forming coalitions for joint decision-making. For each defender partition, the induced allocation game is evaluated by its worst pure-Nash-equilibrium loss together with a supermodular coordination cost. We prove that optimal protection is NP-hard even in a restricted setting, that every fixed partition induces an exact potential game with finite best-response convergence, and that decentralized equilibria can have an unbounded price of anarchy. We further characterize merge and split thresholds, the piecewise-constant dependence of optimal partitions on the coordination weight, the monotonicity of the selected coordination cost, and conditions for diminishing net merge benefits. Based on these properties, we propose Cost-Aware Coalition Search (CACS), which combines a tractable allocation oracle with multi-start local search over merge, split, and move operations. Experiments on synthetic and real networks show that selective coalitions provide a scalable tradeoff between residual attack loss and coordination cost across different resource levels, coordination weights, and defender populations. The framework provides a mathematical basis for organizing distributed defenders in networked security environments.