Open SupOne11 opened 1 month ago
We trained 5 models by modifying the random seeds, and then had all 5 models perform inference simultaneously. We then performed k-means clustering on the final landing points of the 30 predicted future trajectories, resulting in 6 clusters. We then calculated the average trajectory for each cluster separately. Actually, I think most papers do it this way.
Thank you for sharing your work. Could you please explain how the ensemble model mentioned in the paper is specifically implemented, in comparison to the single model?