NTK-Guided Implicit Neural Teaching
This addresses efficiency bottlenecks for researchers and practitioners using INRs in domains like image, audio, and 3D reconstruction, though it is incremental as it builds on existing sampling-based strategies.
The paper tackles the high computational cost of fitting high-resolution signals with Implicit Neural Representations (INRs) by proposing NTK-Guided Implicit Neural Teaching (NINT), which accelerates training by dynamically selecting coordinates based on NTK-augmented loss gradients, resulting in nearly halved training time while maintaining or improving representation quality.
Implicit Neural Representations (INRs) parameterize continuous signals via multilayer perceptrons (MLPs), enabling compact, resolution-independent modeling for tasks like image, audio, and 3D reconstruction. However, fitting high-resolution signals demands optimizing over millions of coordinates, incurring prohibitive computational costs. To address it, we propose NTK-Guided Implicit Neural Teaching (NINT), which accelerates training by dynamically selecting coordinates that maximize global functional updates. Leveraging the Neural Tangent Kernel (NTK), NINT scores examples by the norm of their NTK-augmented loss gradients, capturing both fitting errors and heterogeneous leverage (self-influence and cross-coordinate coupling). This dual consideration enables faster convergence compared to existing methods. Through extensive experiments, we demonstrate that NINT significantly reduces training time by nearly half while maintaining or improving representation quality, establishing state-of-the-art acceleration among recent sampling-based strategies.