Search papers, labs, and topics across Lattice.
This work was supported in part by the National Natural Science Foundation of China under Grant 62575039, Grant 62535017, Grant 62575311, and in part by the Hunan Provincial Natural Science Foundation of China under Grant 2025JJ40059. (Jindong Zhao and Yuan Gao contributed equally to this work) (Corresponding author: Shaobo Xia)Jindong Zhao, Yang Xia and Shaobo Xia are with the School of Aeronautic Engineering, Changsha University of Science and Technology, Hunan, China, 410004. (E-mail: zkintom@gmail.com, xiayang0302@gmail.com, shaoboxia2020@gmail.com)Yuan Gao and Sheng Nie are with the Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, China, 100094. (e-mail: gaoyuan222@mails.ucas.ac.cn, niesheng@aircas.ac.cn).Jun Yue is with the School of Automation, Central South University, Changsha 410083, China (e-mail: junyue@csu.edu.cn).Weiwei Sun is with the University of British Columbia. (weiwei.sun3@gmail.com)
CMU Machine Learning2
0
5
Forget manual curation鈥攁ligning policy gradients with a validation set adaptively selects RL training data, leading to more stable LLM training and improved performance.
LiDAR semantic segmentation models can now generalize across drastically different viewpoints thanks to a novel geometric consistency approach that enforces prediction consistency across augmented views.
