Vision Transformer (ViT) has become one of the cornerstones of the computer vision field, demonstrating exceptional performance. However, its inherent high computational complexity and inference latency still pose significant obstacles for deployment in resource-constrained environments. Token pruning, by removing less informative tokens, offers an effective strategy to reduce computational overhead. However, existing pruning methods largely rely on static or local token importance scores. This myopic approach fundamentally overlooks the sequential dependency of pruning decisions and fails to capture the interaction effects between pruning decisions across layers, often neglecting the global interactions between mask variables. To address this limitation, we propose V-Pruner, a fast and globally-informed token pruning framework for Vision Transformer. V-Pruner first leverages Fisher information to perform an initial assessment of token importance, providing a principled initial prior for pruning decisions. Building on this, V-Pruner introduces a Reinforcement Learning (RL) Proximal Policy Optimization (PPO) algorithm, refining token pruning into a global sequential decision process. The algorithm combines a composite reward signal that incorporates both model performance and computational cost to guide policy exploration, effectively evaluating the long-term impact of different pruning decision combinations on global model performance. Extensive experiments on ViT-L, DeiT-B, DeiT-S, and DeiT-T demonstrate that V-Pruner achieves a better balance between accuracy, GFLOPs, inference speed, and training time, surpassing existing mainstream ViT pruning algorithms in overall performance.

In the crowd sensing application, a slice-based trajectory privacy protection method (STPP) is proposed for the privacy of mobile object trajectory data with rich spatiotemporal information. The trajectory data is first sliced with the position (GPS) as the smallest tuple so that the trajectory is converted into a corresponding trajectory tuple set, and each trajectory tuple is attached with a unique identifier. The user generates three random numbers (α, λ, β) for each trajectory tuple, and sends the trajectory tuple together with (α, λ, β) to the next user who then sends the trajectory tuple and the changed three random numbers to another randomly selected user. In this way, when the threshold of the set random number is reached, the user uploads the trajectory tuple to the server. Thus, the attacker cannot distinguish the location of the user. Due to the loss of the clustering relationship, the uploaded trajectory tuple becomes a set of anonymous and noncontiguous data sets, so that the server can not recognize the individual motion trajectory at all. Finally, the user can reconstruct the trajectory on the client based on the unique identifier. The experimental results and theoretical analysis show that STPP protects the privacy of the user's trajectory and improves the practicability of the data.