Early Termination
Early termination research focuses on efficiently stopping processes before completion, optimizing resource usage and improving performance. Current efforts concentrate on developing algorithms and models, such as those employing reinforcement learning, that predict successful completion or identify points of diminishing returns, enabling early exit strategies in diverse applications like neural network inference and automated machine learning. This research significantly impacts fields ranging from robotics and AI to database management and computer systems optimization by reducing computational costs and improving efficiency.
Papers
Temporal Decisions: Leveraging Temporal Correlation for Efficient Decisions in Early Exit Neural Networks
Max Sponner, Lorenzo Servadei, Bernd Waschneck, Robert Wille, Akash Kumar
Verification-Aided Learning of Neural Network Barrier Functions with Termination Guarantees
Shaoru Chen, Lekan Molu, Mahyar Fazlyab