Polynomial Implicit Neural Representations For Large Diverse Datasets

CVPR 2023  ·  Rajhans Singh, Ankita Shukla, Pavan Turaga ·

Implicit neural representations (INR) have gained significant popularity for signal and image representation for many end-tasks, such as superresolution, 3D modeling, and more. Most INR architectures rely on sinusoidal positional encoding, which accounts for high-frequency information in data. However, the finite encoding size restricts the model's representational power. Higher representational power is needed to go from representing a single given image to representing large and diverse datasets. Our approach addresses this gap by representing an image with a polynomial function and eliminates the need for positional encodings. Therefore, to achieve a progressively higher degree of polynomial representation, we use element-wise multiplications between features and affine-transformed coordinate locations after every ReLU layer. The proposed method is evaluated qualitatively and quantitatively on large datasets like ImageNet. The proposed Poly-INR model performs comparably to state-of-the-art generative models without any convolution, normalization, or self-attention layers, and with far fewer trainable parameters. With much fewer training parameters and higher representative power, our approach paves the way for broader adoption of INR models for generative modeling tasks in complex domains. The code is available at \url{https://github.com/Rajhans0/Poly_INR}

PDF Abstract CVPR 2023 PDF CVPR 2023 Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Image Generation FFHQ 256 x 256 Poly-INR FID 2.72 # 4
Image Generation ImageNet 128x128 Poly-INR FID 2.08 # 3
Image Generation ImageNet 256x256 Poly-INR FID 2.86 # 11
Image Generation ImageNet 512x512 Poly-INR FID 3.81 # 17

Methods


No methods listed for this paper. Add relevant methods here