Generalizable Implicit Neural Representations via Instance Pattern Composers

TL;DR

This paper proposes a novel framework for implicit neural representations that enables a coordinate-based MLP to generalize across data instances by modulating a small set of weights, improving performance on diverse domains.

Contribution

It introduces a simple, effective method for generalizable INRs using weight modulation, compatible with meta-learning and hypernetworks, enhancing cross-instance generalization.

Findings

Achieves high performance on audio, image, and 3D object data.

Validates the effectiveness of weight modulation through ablation studies.

Demonstrates compatibility with existing meta-learning and hypernetwork approaches.

Abstract

Despite recent advances in implicit neural representations (INRs), it remains challenging for a coordinate-based multi-layer perceptron (MLP) of INRs to learn a common representation across data instances and generalize it for unseen instances. In this work, we introduce a simple yet effective framework for generalizable INRs that enables a coordinate-based MLP to represent complex data instances by modulating only a small set of weights in an early MLP layer as an instance pattern composer; the remaining MLP weights learn pattern composition rules for common representations across instances. Our generalizable INR framework is fully compatible with existing meta-learning and hypernetworks in learning to predict the modulated weight for unseen instances. Extensive experiments demonstrate that our method achieves high performance on a wide range of domains such as an audio, image, and 3D object, while the ablation study validates our weight modulation.