ICML 2022spotlight26 citations

PINs: Progressive Implicit Networks for Multi-Scale Neural Representations

Zoe Landgraf, Alexander Sorkine Hornung, Ricardo S Cabral

Abstract

Multi-layer perceptrons (MLP) have proven to be effective scene encoders when combined with higher-dimensional projections of the input, commonly referred to as positional encoding. However, scenes with a wide frequency spectrum remain a challenge: choosing high frequencies for positional encoding introduces noise in low structure areas, while low frequencies results in poor fitting of detailed regions. To address this, we propose a progressive positional encoding, exposing a hierarchical MLP structure to incremental sets of frequency encodings. Our model accurately reconstructs scenes with wide frequency bands and learns a scene representation at progressive level of detail without explicit per-level supervision. The architecture is modular: each level encodes a continuous implicit representation that can be leveraged separately for its respective resolution, meaning a smaller network for coarser reconstructions. Experiments on several 2D and 3D datasets shows improvements in reconstruction accuracy, representational capacity and training speed compared to baselines.

BibTeX
@InProceedings{pmlr-v162-landgraf22a,
  title = 	 {{PIN}s: Progressive Implicit Networks for Multi-Scale Neural Representations},
  author =       {Landgraf, Zoe and Hornung, Alexander Sorkine and Cabral, Ricardo S},
  booktitle = 	 {Proceedings of the 39th International Conference on Machine Learning},
  pages = 	 {11969--11984},
  year = 	 {2022},
  editor = 	 {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan},
  volume = 	 {162},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {17--23 Jul},
  publisher =    {PMLR},
  pdf = 	 {https://proceedings.mlr.press/v162/landgraf22a/landgraf22a.pdf},
  url = 	 {https://proceedings.mlr.press/v162/landgraf22a.html},
  abstract = 	 {Multi-layer perceptrons (MLP) have proven to be effective scene encoders when combined with higher-dimensional projections of the input, commonly referred to as positional encoding. However, scenes with a wide frequency spectrum remain a challenge: choosing high frequencies for positional encoding introduces noise in low structure areas, while low frequencies results in poor fitting of detailed regions. To address this, we propose a progressive positional encoding, exposing a hierarchical MLP structure to incremental sets of frequency encodings. Our model accurately reconstructs scenes with wide frequency bands and learns a scene representation at progressive level of detail without explicit per-level supervision. The architecture is modular: each level encodes a continuous implicit representation that can be leveraged separately for its respective resolution, meaning a smaller network for coarser reconstructions. Experiments on several 2D and 3D datasets shows improvements in reconstruction accuracy, representational capacity and training speed compared to baselines.}
}
PINs: Progressive Implicit Networks for Multi-Scale Neural Representations · ICML 2022