paper-with-me

Papers

A Generalized Spectral Framework to Expain Neural Scaling and Compression Dynamics

2025-11-11 · Yizhou Zhang arxiv

Empirical scaling laws describe how test loss and other performance metrics depend on model size, dataset size, and compute. While such laws are consistent within specific regimes, apparently distinct scaling behaviors have been reported for related settings such as model compression. Motivated by recent progress in spectral analyses of neural representations, this paper develops a \emph{generalized spectral framework} that unifies learning dynamics and compression phenomena under a common functional ansatz. We generalize the spectral evolution function from the linear kernel form $g(λt)=λt$ to an asymptotically polynomial function $g(λ,t;β)$, characterized by an effective spectral--temporal elasticity $ρ(β)$. This framework recovers existing lazy and feature-learning theories as special cases and yields an invariant relation between learning and compression

📄 PDF Abstract BibTeX arXiv:2511.07892

Code (0)

등록된 구현이 없습니다.

Tasks

Model Compression

Similar Papers 제목 키워드 기반

Free Decompression with Algebraic Spectral Curves

2026-05-05 · Siavash Ameli, Chris van der Heide, Liam Hodgkinson, Michael W. Mahoney arxiv

Tools from random matrix theory have become central to deep learning theory, using spectral information to provide mechanisms for modeling generalization, robustness, scaling, and failure modes. While often capable of mo…

Expanded Adaptive Scaling Normalization for End to End Image Compression

2022-08-05 · Chajin Shin, Hyeongmin Lee, Hanbin Son, Sangjin Lee 외

Recently, learning-based image compression methods that utilize convolutional neural layers have been developed rapidly. Rescaling modules such as batch normalization which are often used in convolutional neural networks…

Image Compression

TexPainter: Generative Mesh Texturing with Multi-view Consistency

2024-05-17 · HongKun Zhang, Zherong Pan, Congyi Zhang, Lifeng Zhu 외

The recent success of pre-trained diffusion models unlocks the possibility of the automatic generation of textures for arbitrary 3D meshes in the wild. However, these models are trained in the screen space, while convert…

Denoising

FlexPainter: Flexible and Multi-View Consistent Texture Generation

2025-06-03 · Dongyu Yan, Leyi Wu, Jiantao Lin, Luozhou Wang 외

Texture map production is an important part of 3D modeling and determines the rendering quality. Recently, diffusion-based methods have opened a new way for texture generation. However, restricted control flexibility and…

Texture Synthesis

Generalized Inpainting Method for Hyperspectral Image Acquisition

2015-02-06 · K. Degraux, V. Cambareri, L. Jacques, B. Geelen 외

A recently designed hyperspectral imaging device enables multiplexed acquisition of an entire data volume in a single snapshot thanks to monolithically-integrated spectral filters. Such an agile imaging technique comes a…

compressed sensingDemosaicking