Tag
This paper investigates the role of parameter symmetry in weight-space perception, showing that symmetry scatter alone accounts for nearly the entire degradation in performance between shared and independently-fitted neural networks. The study uses SIRENs and large-scale experiments to argue that computational advantages may justify weight-space methods over informational equivalence to function access.
This paper introduces a recurrent sinusoidal architecture for implicit neural representations that achieves higher fidelity with fewer parameters and optimization steps by exploiting harmonic line spectrum enrichment through sinusoidal recurrence.
This paper presents a video codec that stores video and audio as weights of a sinusoidal representation network, using knowledge distillation and quantization for compression. Experiments show a 2.61x compression ratio compared to the original network, but quality lags behind standard codecs like H.264 and HEVC.
The paper proposes a lightweight method that reformulates regression-based INR training as a classification task by discretizing continuous targets into bins, enabling flexible distribution modeling for error-aware uncertainty estimation in scientific data compression.
This paper introduces a differentiable training objective based on the Rice level-crossing density from random-field theory, which acts as a mesh-free high-frequency auxiliary loss for implicit neural representations. It requires no grid or FFT, and improves performance on non-uniform samples.
This paper studies transfer specificity in implicit neural representations across SIREN, ReLU MLPs, and Fourier-feature MLPs, finding that transfer magnitude and specificity depend on architecture, with ReLU being more selective and SIREN reusing weights broadly. Results suggest architecture selection should consider explicit control conditions, not just transfer magnitude.
This paper presents a comprehensive taxonomy of 3D vision research, covering geometric representations, datasets, learning paradigms, and applications in reconstruction, generation, and video modeling.
This paper introduces the Spectral Energy Centroid (SEC) metric to analyze and improve spectral bias in implicit neural representations, demonstrating its utility for hyperparameter selection, signal complexity measurement, and cross-architecture alignment.