GitHub

Reward-Hacking-Resistant RL Environment for ML Systems

The solution batches all tokens routed to the same expert into a single matrix multiplication call instead of looping over each token individually. This is the key reason why speedup increases with ...
GitHub

Reward-Hacking-Resistant RL Environment for ML Systems

End-to-end RL environment design: sparse MoE task, tamper-resistant judge, and PyTorch performance analysis inspired by ScatterMoE - ji24077/Reward-Hacking-Resistant-RL-Environment-for-ML-Systems ...
IEEE

Adaptive Dynamic Window Approach for Mobile Robot Motion Control Using Particle Swarm Optimization

Abstract: Autonomous navigation for mobile robots in dynamic and unknown environments requires a robust and adaptable path planning approach able to handle real world various applications. Among the ...
IEEE

Adaptive Dynamic Functional Connectivity with Deep Spatio-Temporal Fusion for High-Accuracy Identification of Major Depressive Disorder

Abstract: The pathogenesis of major depressive disorder (MDD) has not been fully elucidated, and early identification and intervention are the most effective approach. Dynamic functional connectivity ...