Beyond Heuristics: Globally Optimal Configuration of Implicit Neural Representations
This addresses a foundational bottleneck in INR design for researchers and practitioners in signal processing and computer vision, replacing ad-hoc heuristics with a systematic approach.
The paper tackles the problem of optimal configuration for Implicit Neural Representations (INRs), which lack principled strategies, by introducing OptiINR, a unified framework that formulates this as an optimization problem using Bayesian optimization, resulting in globally optimal configurations that maximize performance across diverse applications.
Implicit Neural Representations (INRs) have emerged as a transformative paradigm in signal processing and computer vision, excelling in tasks from image reconstruction to 3D shape modeling. Yet their effectiveness is fundamentally limited by the absence of principled strategies for optimal configuration - spanning activation selection, initialization scales, layer-wise adaptation, and their intricate interdependencies. These choices dictate performance, stability, and generalization, but current practice relies on ad-hoc heuristics, brute-force grid searches, or task-specific tuning, often leading to inconsistent results across modalities. This work introduces OptiINR, the first unified framework that formulates INR configuration as a rigorous optimization problem. Leveraging Bayesian optimization, OptiINR efficiently explores the joint space of discrete activation families - such as sinusoidal (SIREN), wavelet-based (WIRE), and variable-periodic (FINER) - and their associated continuous initialization parameters. This systematic approach replaces fragmented manual tuning with a coherent, data-driven optimization process. By delivering globally optimal configurations, OptiINR establishes a principled foundation for INR design, consistently maximizing performance across diverse signal processing applications.