arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2209.01019cs.CV
收藏

论隐式神经表示的量化

On Quantizing Implicit Neural Representations

University of Adelaide(阿德莱德大学)

Cameron Gordon, Shin-Fang Chng, Lachlan MacDonald, Simon Lucey

AI总结:

针对隐式神经网络在低比特率下量化效果差的问题,提出非均匀聚类量化方法,并通过架构搜索将NeRF压缩至16kb以下且性能损失极小。

摘要:

量化在隐式/坐标神经网络中的作用仍未被完全理解。我们注意到,在训练期间使用规范的固定量化方案,由于网络权重分布在训练过程中不断变化,会导致在低比特率下表现不佳。在本研究中,我们展示了神经权重的非均匀量化能够带来显著的性能提升。具体而言,我们证明了聚类量化能够改善重建效果。最后,通过刻画量化与网络容量之间的权衡关系,我们证明了使用二值神经网络重建信号是可能的(尽管内存效率较低)。我们在2D图像重建和3D辐射场上通过实验验证了我们的发现;并展示了简单的量化方法和架构搜索能将NeRF压缩至16kb以下,且性能损失极小(比原始NeRF小323倍)。

英文摘要:

The role of quantization within implicit/coordinate neural networks is still not fully understood. We note that using a canonical fixed quantization scheme during training produces poor performance at low-rates due to the network weight distributions changing over the course of training. In this work, we show that a non-uniform quantization of neural weights can lead to significant improvements. Specifically, we demonstrate that a clustered quantization enables improved reconstruction. Finally, by characterising a trade-off between quantization and network capacity, we demonstrate that it is possible (while memory inefficient) to reconstruct signals using binary neural networks. We demonstrate our findings experimentally on 2D image reconstruction and 3D radiance fields; and show that simple quantization methods and architecture search can achieve compression of NeRF to less than 16kb with minimal loss in performance (323x smaller than the original NeRF).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑