FINER: Flexible spectral-bias tuning in Implicit NEural Representation by Variable-periodic Activation Functions
This addresses a bottleneck in INR for signal processing, offering a novel method to enhance representation of multi-frequency signals, though it appears incremental as it builds on existing INR techniques.
The paper tackles the problem of limited frequency tuning in Implicit Neural Representations (INR) for representing complex signals, proposing FINER with variable-periodic activation functions to flexibly tune the supported frequency set, resulting in improved performance in tasks like 2D image fitting, 3D signed distance fields, and 5D neural radiance fields, outperforming existing INRs.
Implicit Neural Representation (INR), which utilizes a neural network to map coordinate inputs to corresponding attributes, is causing a revolution in the field of signal processing. However, current INR techniques suffer from a restricted capability to tune their supported frequency set, resulting in imperfect performance when representing complex signals with multiple frequencies. We have identified that this frequency-related problem can be greatly alleviated by introducing variable-periodic activation functions, for which we propose FINER. By initializing the bias of the neural network within different ranges, sub-functions with various frequencies in the variable-periodic function are selected for activation. Consequently, the supported frequency set of FINER can be flexibly tuned, leading to improved performance in signal representation. We demonstrate the capabilities of FINER in the contexts of 2D image fitting, 3D signed distance field representation, and 5D neural radiance fields optimization, and we show that it outperforms existing INRs.