‘Improving Equivariant Model Training via Constraint Relaxation’, NeurIPS 2024: Introduced a method to improve equivariant model training by relaxing constraints during training.
‘BiEquiFormer: Bi-Equivariant Representations for Global Point Cloud Registration’, NeurReps 2024: Proposed a bi-equivariant representation-based method for pose-invariant point cloud registration.
‘SE(3)-Equivariant Attention Networks for Shape Reconstruction in Function Space’, ICLR 2023: Developed an SE(3)-equivariant Transformer for shape reconstruction from point clouds.
‘Learning Augmentation Distributions using Transformed Risk Minimization’, TMLR 2023: Proposed Transformed Risk Minimization (TRM) to jointly learn models and augmentation distributions.
‘Detecting Adversarial Examples in Convolutional Neural Networks’, arXiv 2018: Investigated techniques for detecting adversarial examples in CNNs.