AI Paper Atlas

Research map / 3D vision / Face and shape

Scaling Optimization-Oriented Hypernetworks for Implicit Neural Representations

Lulu Cai, Hanxiang Ren, Siyan Dong, Li Sun, Jiefeng Wu, Youyi Zheng, Yi Ma, Yanchao Yang

NeurIPS 2026 · 3D vision › Face and shape · approach: Implicit neural representation design · 0 citations (OpenAlex)

Read the paper

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a hypernetwork that generates INR parameters for a signal in one forward pass, preserving the chain-rule structure of test-time optimization while scaling better than prior quartic-parameter designs.

Problem

Generating implicit neural representation (INR) parameters for a given signal without test-time optimization; existing hypernetworks either ignore the chain-rule structure of test-time optimization or instantiate it with MLPs whose parameter count grows quartically in target size.

Approach

A hypernetwork that mirrors the chain-rule structure of optimization-based INR fitting, so that the generated parameters follow an optimization-oriented update, while avoiding the quartic parameter growth of prior MLP instantiations.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
tokenize weights by input neuron (column) instead of output neuron (row) (ablation)hypernetwork · CelebA-HQPSNR (dB): 25.3 -> 27.7Improved
use output-neuron-centric tokenization (ablation)hypernetwork · ImageNetPSNR (dB): 19.8 -> 22.9Improved
fix cross-attention QKV roles by information-flow consistency (ablation)hypernetwork · NeRF (weights & images)PSNR (dB): 22.9 -> 27.8Improved
fix QKV roles by information-flow consistency (ablation)hypernetwork · CelebA-HQPSNR (dB): 23.5 -> 27.7Improved
replace MLP body with attention in optimization-oriented hypernetworkOurs vs HyPoGen · ImageNetPSNR (dB): 13.2 -> 22.9Improved
add optimization-oriented structure to attention hypernetworkOurs vs LDMI · CelebA-HQPSNR (dB): 24.8 -> 27.7Improved
add optimization-oriented structure under weights-only supervisionOurs vs LDMI · Pre-trained NeRF (weights only)PSNR (dB): 19.3 -> 22.0Improved
MLP-based optimization-oriented body vs attention-only body (HyPoGen vs LDMI)HyPoGen · Pre-trained NeRF (weights only)PSNR (dB): 19.9 vs 19.3Mixed
train without ground-truth INR weights (images only)Ours · Pre-trained NeRFPSNR (dB): 27.8 -> 28.8Improved
increase NeRF training data from 10% to 100% (ablation)Ours · Pre-trained NeRFPSNR (dB): +3.2 dB (LDMI +2.5, HyPoGen +1.8)Improved
increase target INR width from 128 to 1024 (ablation)Ours · CelebA-HQPSNR (dB): 26.5 -> 28.8Improved
increase target INR width to 1024 (ablation)HyPoGen · CelebA-HQPSNR (dB)Hurt

Experimental setup

Builds on: optimization-oriented hypernetworks

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy