Targeted multi-agent communication algorithm based on state control

As an important mechanism in multi-agent interaction,communication can make agents form complex team relationships rather than constitute a simple set of multiple independent agents.However,the existing communication schemes can bring much timing redundancy and irrelevant messages,which seriously affects their practical application.To solve this problem,this paper proposes a targeted multiagent communication algorithm based on state control(SCTC).The SCTC uses a gating mechanism based on state control to reduce the timing redundancy of communication between agents and determines the interaction relationship between agents and the importance weight of a communication message through a series connection of hard-and self-attention mechanisms,realizing targeted communication message processing.In addition,by minimizing the difference between the fusion message generated from a real communication message of each agent and a fusion message generated from the buffered message,the correctness of the final action choice of the agent is ensured.Our evaluation using a challenging set of Star Craft II benchmarks indicates that the SCTC can significantly improve the learning performance and reduce the communication overhead between agents,thus ensuring better cooperation between agents.

Same same but different:Neuroanatomic and connectomic profiles of early-onset schizophrenia and autism spectrum disorder

Background:Early-onset schizophrenia(EOS)and autism spectrum disorder(ASD)are two mental disorders with abnormal neurodevelopment and overlapping clinical symptoms.However,few prospective multi-center imaging studies explored their pathophysiological differences and similarities comprehensively.Methods:We utilized a large cohort of functional magnetic resonance imaging data with standardized scanning protocols from 594 participants(276 participants with high-functioning ASD,187 participants with EOS,and 131 normal controls(NCs))recruited at five centers.A comprehensive evaluation of anatomical and connectomic features,in-cluding cortical thickness,subcortical nuclei volume,functional connectivity,functional gradients,and network efficiency,was conducted,culminating in the construction of imaging-based neuro-profiles for individuals with EOS and ASD.Results:The derived neuro-profiles revealed that EOS is characterized by more pro-nounced reductions in cortical thickness and subcortical nuclei volumes.In contrast,ASD exhibited more pronounced abnormalities in connectomic attributes.The de-creased network functional connectivity and compressed functional gradients in ASD are likely to contribute to reduced network efficiency.Classifiers based on imaging profiles successfully distinguished individuals with EOS,ASD,and NCs with com-mendable performance in leave-site-out cross-validation(area under the operating characteristic curve(AUC)for ASD vs.NC:0.769,AUC for ASD vs.EOS:0.792,AUC for EOS vs.NC:0.854).Conclusion:The neuro-profiles revealed that EOS and ASD exhibit overlapping as well as unique imaging characteristics;however,EOS exhibits greater structural variation,while ASD presents with more subtle connectomic abnormalities.These findings offer valuable insights into the distinct neurobiological mechanisms of EOS and ASD and enhance the prospect of early diagnosis through objective imaging-based biomarkers.