Predicting the Encoding Error of SIRENs
Not on the map. Why?
Abstract (arXiv)
Implicit Neural Representations (INRs), which encode signals such as images, videos, and 3D shapes in the weights of neural networks, are becoming increasingly popular. Among their many applications is signal compression, for which there is great interest in achieving the highest possible fidelity to the original signal subject to constraints such as neural network size, training (encoding) and inference (decoding) time. But training INRs can be a computationally expensive process, making it challenging to determine the best possible tradeoff under such constraints. Towards this goal, we present a method which predicts the encoding error that a popular INR network (SIREN) will reach, given its network hyperparameters and the signal to encode. This method is trained on a unique dataset of 300,000 SIRENs, trained across a variety of images and hyperparameters. (Dataset available here: https://huggingface.co/datasets/predict-SIREN-PSNR/COIN-collection.) Our predictive method demonstrates the feasibility of this regression problem, and allows users to anticipate the encoding error that a SIREN network will reach in milliseconds instead of minutes or longer. We also provide insights into the behavior of SIREN networks, such as why narrow SIRENs can have very high random variation in encoding error, and how the performance of SIRENs relates to JPEG compression.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 53 of its 64 references that are in the map. First posted on arXiv 29 Oct 2024.
Builds on
53 of 64 references.
References · 53
- 2014Microsoft COCO: Common Objects in Contextcited by 17,608
- 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
- 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,723
- 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,621
- 2020Implicit Neural Representations with Periodic Activation Functionscited by 2,238
- 2020Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domainscited by 2,200
- 2020D-NeRF: Neural Radiance Fields for Dynamic Scenescited by 1,465
- 2020Nerfies: Deformable Neural Radiance Fieldscited by 1,198
- 2016HyperNetworkscited by 1,027
- 2023LRM: Large Reconstruction Model for Single Image to 3Dcited by 849
- 2020Neural Scene Flow Fields for Space-Time View Synthesis of Dynamic Scenescited by 751
- 2019Fine-Grained Analysis of Optimization and Generalization for Overparameterized Two-Layer Neural Networkscited by 566
38 more
- 2020Learning Continuous Image Representation with Local Implicit Image Functioncited by 552
- 2020Non-Rigid Neural Radiance Fields: Reconstruction and Novel View Synthesis of a Dynamic Scene From Monocular Videocited by 471
- 2021Dynamic View Synthesis from Dynamic Monocular Videocited by 420
- 2020Space-time Neural Irradiance Fields for Free-Viewpoint Videocited by 414
- 2021Explaining Neural Scaling Lawscited by 363
- 2021High-Performance Large-Scale Image Recognition Without Normalizationcited by 360
- 2021NeRV: Neural Representations for Videoscited by 256
- 2020Learned Initializations for Optimizing Coordinate-Based Neural Representationscited by 242
- 2021COIN: COmpression with Implicit Neural representationscited by 233
- 2023WIRE: Wavelet Implicit Neural Representationscited by 220
- 2022From data to functa: Your data point is a function and you can treat it like onecited by 187
- 2021Implicit Neural Representations for Image Compressioncited by 162
- 2021BACON: Band-limited Coordinate Networks for Multiscale Scene Representationcited by 143
- 2021Modulated Periodic Activations for Generalizable Local Functional Representationscited by 140
- 2019The Convergence Rate of Neural Networks for Learned Functions of Different Frequenciescited by 135
- 2021Beyond Periodicity: Towards a Unifying Framework for Activations in Coordinate-MLPscited by 135
- 2021A Structured Dictionary Perspective on Implicit Neural Representationscited by 100
- 2023ATT3D: Amortized Text-to-3D Object Synthesiscited by 87
- 2022COIN++: Neural Compression Across Modalitiescited by 82
- 2022E-NeRV: Expedite Neural Video Representation with Disentangled Spatial-Temporal Contextcited by 76
- 2021Implicit Neural Video Compressioncited by 52
- 2022NIRVANA: Neural Implicit Representations of Videos with Adaptive Networks and Autoregressive Patch-wise Modelingcited by 51
- 2021Seeing Implicit Neural Representations as Fourier Seriescited by 50
- 2021Meta-Learning Sparse Implicit Neural Representationscited by 48
- 2024LATTE3D: Large-scale Amortized Text-To-Enhanced3D Synthesiscited by 39
- 2023Towards Scalable Neural Representation for Diverse Videoscited by 37
- 2022A Fast, Well-Founded Approximation to the Empirical Neural Tangent Kernelcited by 30
- 2022Shape, Pose, and Appearance from a Single Image via Bootstrapped Radiance Field Inversioncited by 30
- 2022Scalable Neural Video Representations with Learnable Positional Featurescited by 29
- 2023Disorder-invariant Implicit Neural Representationcited by 27
- 2022The Onset of Variance-Limited Behavior for Networks in the Lazy and Rich Regimescited by 25
- 2023Compression with Bayesian Implicit Neural Representationscited by 25
- 2020Predicting Training Time Without Trainingcited by 22
- 2022Meta-Learning Sparse Compression Networkscited by 22
- 2022CNeRV: Content-adaptive Neural Representation for Visual Datacited by 21
- 2022On Quantizing Implicit Neural Representationscited by 20
- 2022Neural Residual Flow Fields for Efficient Video Representationscited by 16
- 2018Efficient Image Dataset Classification Difficulty Estimation for Predicting Deep-Learning Accuracycited by 14
50 most cited of 53 All 53 (JSON)
Built on by
Citing papers per year: 2025: 2.
2 papers.