To be updated.
Investigating how different matching strategies affect the user experience of ride-sharing services through systematic simulation.
Enhance the practicality of a prediction model by improving computing efficiency and the accuracy of the upgraded prediction model through simulation. Aiming to provide operational insights by analyzing simulation results thereby improving the practice of the prediction model, and contributing to the improvement of ride-sharing services.
A system was designed to autonomously detect small orbital remnants (less than 10*10 pixels on the image) through unmanned patrol vehicles and send warnings accordingly. The detection system was based on Yolo V3.
Combined meta-learning and Hyperparameter Optimization via a double-layer programming structure. Verified that the approach of ‘approximate gradient’ can greatly reduce the computational time complexity and the computational efficiency compared to the ‘gradient descent’ hyperparameter optimization method.