From Theory to Throughput: CUDA-Optimized APML for Large-Batch 3D Learning
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866911333315772416 |
|---|---|
| author | Sharifipour, Sasan Casado, Constantino Álvarez Cañellas, Manuel Lage López, Miguel Bordallo |
| author_facet | Sharifipour, Sasan Casado, Constantino Álvarez Cañellas, Manuel Lage López, Miguel Bordallo |
| contents | Loss functions are fundamental to learning accurate 3D point cloud models, yet common choices trade geometric fidelity for computational cost. Chamfer Distance is efficient but permits many-to-one correspondences, while Earth Mover Distance better reflects one-to-one transport at high computational cost. APML approximates transport with differentiable Sinkhorn iterations and an analytically derived temperature, but its dense formulation scales quadratically in memory. We present CUDA-APML, a sparse GPU implementation that thresholds negligible assignments and runs adaptive softmax, bidirectional symmetrization, and Sinkhorn normalization directly in COO form. This yields near-linear memory scaling and preserves gradients on the stored support, while pairwise distance evaluation remains quadratic in the current implementation. On ShapeNet and MM-Fi, CUDA-APML matches dense APML within a small tolerance while reducing peak GPU memory by 99.9%. Code available at: https://github.com/Multimodal-Sensing-Lab/apml |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_19743 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | From Theory to Throughput: CUDA-Optimized APML for Large-Batch 3D Learning Sharifipour, Sasan Casado, Constantino Álvarez Cañellas, Manuel Lage López, Miguel Bordallo Machine Learning Artificial Intelligence Computational Geometry Loss functions are fundamental to learning accurate 3D point cloud models, yet common choices trade geometric fidelity for computational cost. Chamfer Distance is efficient but permits many-to-one correspondences, while Earth Mover Distance better reflects one-to-one transport at high computational cost. APML approximates transport with differentiable Sinkhorn iterations and an analytically derived temperature, but its dense formulation scales quadratically in memory. We present CUDA-APML, a sparse GPU implementation that thresholds negligible assignments and runs adaptive softmax, bidirectional symmetrization, and Sinkhorn normalization directly in COO form. This yields near-linear memory scaling and preserves gradients on the stored support, while pairwise distance evaluation remains quadratic in the current implementation. On ShapeNet and MM-Fi, CUDA-APML matches dense APML within a small tolerance while reducing peak GPU memory by 99.9%. Code available at: https://github.com/Multimodal-Sensing-Lab/apml |
| title | From Theory to Throughput: CUDA-Optimized APML for Large-Batch 3D Learning |
| topic | Machine Learning Artificial Intelligence Computational Geometry |
| url | https://arxiv.org/abs/2512.19743 |