SAPE: Spatially-Adaptive Progressive Encoding for Neural Optimization
This addresses a fundamental limitation in neural optimization for researchers and practitioners in fields like computer vision and graphics, though it appears incremental as it builds on existing encoding methods.
The paper tackles the problem of multilayer perceptrons struggling to learn functions with wide frequency bands by introducing a spatially adaptive progressive encoding (SAPE) scheme, which improves fitting across frequencies without sacrificing training stability or requiring preprocessing, as demonstrated in tasks like signal regression and 3D mesh transfer.
Multilayer-perceptrons (MLP) are known to struggle with learning functions of high-frequencies, and in particular cases with wide frequency bands. We present a spatially adaptive progressive encoding (SAPE) scheme for input signals of MLP networks, which enables them to better fit a wide range of frequencies without sacrificing training stability or requiring any domain specific preprocessing. SAPE gradually unmasks signal components with increasing frequencies as a function of time and space. The progressive exposure of frequencies is monitored by a feedback loop throughout the neural optimization process, allowing changes to propagate at different rates among local spatial portions of the signal space. We demonstrate the advantage of SAPE on a variety of domains and applications, including regression of low dimensional signals and images, representation learning of occupancy networks, and a geometric task of mesh transfer between 3D shapes.