Self Influence

Self-influence, encompassing how an agent's actions affect its own future states and outcomes, is a burgeoning research area with applications across robotics, AI, and human-computer interaction. Current research focuses on improving decision-making in complex, multi-agent systems, often employing techniques like variational autoencoders to model uncertain behaviors and influence vectors to control robotic actuators. These studies aim to enhance robustness, efficiency, and explainability in various applications, particularly by improving the reliability of influence function methods for data cleaning and model interpretation in machine learning. The ultimate goal is to create more adaptable, reliable, and human-centered systems.