NPNet: A Non-Parametric Network with Adaptive Gaussian–Fourier Positional Encoding for 3D Classification and Segmentation
Mohammad Saeid, Amir Salarpour, Pedram MohajerAnsari, Mert D. Pesé · 2026
Read NPNet: A Non-Parametric Network with Adaptive Gaussian–Fourier Positional Encoding for 3D Classification and Segmentation by Mohammad Saeid, Amir Salarpour, Pedram MohajerAnsari, Mert D. Pesé (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-00542.txt.
Description
Mohammad Saeid and colleagues present NPNet, a non-parametric method for classifying and segmenting 3D point clouds with no learned weights. Point features come from fixed operators: farthest point sampling, k-nearest neighbours, and pooling. Its adaptive Gaussian-Fourier encoding picks bandwidth and mixing from geometry, staying stable across scales and sampling densities, with Fourier features adding context for segmentation. On test sets it performs strongly among non-parametric baselines, especially few-shot on ModelNet40, with modest memory use and inference time.
Catalog record
| CREATOR | Mohammad Saeid, Amir Salarpour, Pedram MohajerAnsari, Mert D. Pesé |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 30954 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602.00542