Bidirectional Mamba Enhances Trusted Collaborator Selection.

Researchers have introduced a novel approach, the Bidirectional Mamba (BM) model, designed to improve the selection of trustworthy collaborators in distributed systems. The core challenge lies in accurately assessing long-term device behavior, which is crucial for predicting future reliability in collaborative tasks. Traditional methods often fall short by only capturing instantaneous behavior or relying on unidirectional temporal analysis. The BM model addresses these limitations by constructing graphs of device interactions within short time slots and aggregating behavioral features. These short-term representations are then fed into a bidirectional Mamba model, which is capable of integrating information across all time intervals, considering both past and future dependencies. This comprehensive temporal analysis yields a more stable and reliable long-term behavior evaluation for each device. Experimental results indicate that the BM model surpasses baseline methods in evaluation accuracy. This enhanced accuracy directly translates to more effective selection of collaborators, ultimately maximizing the successful completion of collaborative tasks by ensuring that only the most reliable devices are chosen.